Elon Musk’s artificial intelligence startup xAI is planning to build a supercomputer specifically designed to power the next iteration of its AI chatbot Grok, according to a report by The Information.

Musk wants to have the supercomputer operational by Fall 2025, and “will hold himself responsible for delivering it on time,” the report said citing a presentation by xAI to investors in May.

The proposed supercomputer will reportedly utilize 100,000 Nvidia H100 GPUs (graphic processing units) and be at least four times larger than the biggest GPU clusters currently in existence, such as those “used by Meta to train its AI models.” This immense power is necessary to train and run the ever-evolving Grok language model.

The report also added that Musk’s company is partnering with Oracle to build the supercomputer.

Queries to xAI and Oracle remain unanswered.

The reported supercomputer project coincides with xAI’s recent announcement of a $6 billion series B funding round.

“xAI is pleased to announce our Series B funding round of $6 billion with participation from key investors including Valor Equity Partners, Vy Capital, Andreessen Horowitz, Sequoia Capital, Fidelity Management & Research Company, Prince Alwaleed Bin Talal and Kingdom Holding, amongst others,” the company said in a statement.

The funds from the round will be used to take xAI’s first products to market, build advanced infrastructure, and accelerate the research and development of future technologies, the statement added. xAI  also said that it “will continue on this steep trajectory of progress over the coming months, with multiple exciting technology updates and products soon to be announced.”

This funding is expected to fuel xAI’s product development, infrastructure advancements, and future AI research.

Grok’s need for supercomputing power

xAI launched Grok in November 2023 inspired by “The Hitchhiker’s Guide to the Galaxy” and projected as “more than just a traditional chatbot.” It strives to answer user queries in an informative and even witty manner, while also prompting users with relevant questions that they might not have considered. 

However, as Grok evolved through iterations like Grok 1.5 with long-context understanding and Grok 1.5V with visual information processing, its computational requirements have dramatically increased. Training Grok 2 model required 20,000 Nvidia H100 GPUs and Grok 3 and beyond would require 100,000 Nvidia H100 GPUs, the report added.

This exceptional growth in computational requirements necessitates a supercomputer. The current frontrunners in the supercomputer race are Frontier (US), Aurora (US), Eagle (US), and Fugaku (Japan). While the exact specifications of xAI’s planned supercomputer are not known yet, it would fall within the supercomputer category, if xAI’s above claims are met, potentially challenging the dominance of existing machines.

Research by the UK’s Competition and Markets Authority (CMA) has uncovered concerns that cloud users’ established commitments with hyperscalers may limit their future cloud computing options.

Public cloud infrastructure services were referred to the CMA for a market investigation by Ofcom, the communications regulator, which is investigating concerns that technical and business barriers such as software licensing conditions make it difficult for UK enterprises to either make use of multiple cloud service providers or to switch between them.

Factors such as egress fees — which service providers charge customers for extracting data from their cloud — and preliminary discounts might be contributing to vendor lock-in, regulators fear.

These, along with issues such as software licensing, could disincentivise customers from using rival cloud providers, thereby stifling innovation and inhibiting competition in the cloud service market.

Expert panel

The CMA commissioned research among 50 medium to large UK public cloud customers (pdf) who use one of the five largest cloud providers in the UK, Amazon Web Services, Microsoft Azure, Google Cloud, IBM Cloud, and Oracle Cloud. The study focused on perceptions of their infrastructure-as-a-service (IaaS) and platform-as-a-service (PaaS) offerings.

Private clouds and software-as-a-service offerings were considered out of scope.

Switching concerns

Few of those interviewed had chosen to switch between public cloud providers. Switching was motivated by anticipated reductions in costs and promises of improved performance. Those who had switched tended to report that the process was more challenging and time-consuming than anticipated.

Egress fees failed to emerge as a major concern for most participants, even though there were complaints about pricing transparency in this area. Committed spend discounts might also disincentivise switching.

Software licensing practices were not widely viewed, in isolation, as influencing the choice of public cloud provider. Microsoft shops may pick Microsoft Azure, for example, as the most convenient option.

However technical barriers such as a lack of code portability, especially in the context of PaaS and databases, emerged as a bigger barrier to cloud customers considering either a multi-cloud approach or switching cloud providers.

Survey participants were also concerned that as artificial intelligence becomes a more integral component of cloud computing services it will become more difficult to switch providers.

A lack of transparency in public cloud charges emerged as the greatest source of complaint in the survey, which will inform the next steps in the CMA’s investigation.

In response to queries from Network World, AWS said it was committed to “work constructively with the CMA” through its investigation.

“We believe the CMA will recognize how the increased competition, improved security, and lower costs of IT services introduced by cloud technologies benefits UK companies and the overall economy,” an AWS spokesperson said. “AWS designs cloud services to give customers the freedom to choose the technology that best suits their needs, and the cloud has made switching between providers easier than ever.”

Microsoft, which also said it was ready to work with the CMA, said the cloud services market was already competitive.

“The cloud industry is innovative, highly competitive and an accelerator for growth across the UK economy,” it said. “While Amazon, Google and Microsoft’s cloud offerings all continue to grow at a healthy pace, competition is not limited to these providers, and many other companies have thriving cloud businesses.”

Wider perspective

Adrian Bradley, head of cloud transformation at KPMG UK, told Network World that the focus of the CMA’s review was too narrow.

“Hyperscale cloud, traditional IT, and on-premises environments are all important parts of an organisation’s cloud strategy, and enterprises should consider every option,” according to Bradley.

Bradley disagreed that contract terms such as committed spend discounts were inhibiting competition.

“Many organisations benefit from the up-front financial incentives that these providers offer, without which they may not be able to afford digital transformation,” Bradley explained. “At the macro level, these incentives are important to the wider UK technology services economy.”

The CMA investigation needs to keep pace with a fast-evolving market, Bradley concluded.

For example, some major cloud players have stopped charging customers an egress fee to leave their storage platform, according to Bradley.

“If the CMA can effectively navigate the nuances of this already dynamic market, they will create an environment with fair and efficient access to the cloud and the next generation of technologies,” Bradley said.

Running commands in the Linux terminal is likely something you do nearly every day, but moving a running process into the “background” and later moving it back into the “foreground” is something different altogether. When a process is running in the foreground (the usual state of running commands), it could be said to “own your terminal”. In other words, you will likely sit and wait for it to finish before you even try to run another command. If a process is likely to take a long time to complete, on the other hand, you do have an option other than waiting patiently for it to get done. You can move it into the background – freeing up your terminal window to run other commands.

Moving a process to the background

To move a process into the background, you would type ^z (hold the Ctrl key and press “z”) to suspend it, and then type bg to move it into the background. The process will keep running, but you’ll get your prompt back and be able to work on something else while it’s still running to completion in the background. Note that a backgrounded process will keep running until it’s done or until you log off. Here’s an example:

$ sleep 1000
^Z
[1]+  Stopped                 sleep 1000
$ bg
[1]+ sleep 1000 &

Checking on backgrounded processes with the jobs command

To list backgrounded processes, use the jobs command like this:

$ jobs
[1]+  Running                 sleep 1000 &

Note that the backgrounded processes will have a job number (1 in this case).

Moving a process back into the foreground

To move a backgrounded process back into the foreground, use the fg command (if there’s only one) or the fg command followed by the job id to specify a particular process. It will take over your terminal window.

$ fg
sleep 1000

You can then wait for the process to finish or move it again into the background.

Going between foreground and background

In the example scenario below, we start two processes and stop each with a ^z.

$ sleep 1000; sleep 2000

^Z
[1]+  Stopped                 sleep 1000

^Z
[2]+  Stopped                 sleep 2000

We then background both of the processes with a bg command including the job numbers.

$ bg 1; bg 2
[1]- sleep 1000 &
[2]+ sleep 2000 &

$ jobs
[1]-  Running                 sleep 1000 &
[2]+  Running                 sleep 2000 &

The step below shows moving only the second backgrounded process back into the foreground.

$ fg 2
sleep 2000

Running a process in the background from the start

If you want to run a command in the background from the start, add an asterisk to the end of the command (e.g., runme &). This will put the process immediately into the background. After that, you can run whatever other commands you want to run and check on the backgrounded process as needed with the jobs command to see if it’s still running.

When the backgrounded process finishes, you should see something like this:

[1]+  Done             runme

Stopping a backgrounded process

If you want to stop a backgrounded process before you log off, you can bring it back to the foreground with an fg command and then press ^c (Ctrl + “c”) to kill it. Notice that, after doing this, the jobs command has no output:

$ jobs
[1]+  Running                 sleep 1000 &
$ fg 1
sleep 1000
^C
$ jobs
$

Disowning a process

If you want to make sure a process running in the background will keep running even after you log off, you can use a disown command. In the command below, we start a process in the background and display it with the jobs -l command which includes the process ID (3926) rather than just the job number (1) in its output.

$ sleep 1234 &
[1] 3926
$ jobs
[1]+  Running                 sleep 1234 &
$ jobs -l
[1]+  3926 Running                 sleep 1234 &

We then use the disown command to allow it to run outside of our terminal’s control. Note that the jobs command does not display any information on the disowned process.

$ disown 3926
$ jobs

When we log in again, the process might still be running – depending, in this case, on how many seconds of sleep we requested. We can look for it by looking for the process ID or command name:

$ ps -ef | grep sleep
shs         3926       1  0 14:32 ?        00:00:00 sleep 1234
shs         4030    3982  0 14:36 pts/1    00:00:00 grep --color=auto sleep
$ ps -ef | grep 3926
shs         4145    3982  0 14:55 pts/1    00:00:00 grep --color=auto 3926

Wrap-up

Running processes in the background can be convenient when you want to use your terminal window for something else while you’re waiting for the first task to complete. It just takes a little getting used to using &, ^z, bg, fg and the disown commands.

The Federal Communications Commission (FCC) has proposed a hefty $6 million fine against a political consultant for allegedly using AI-generated voice cloning and caller ID spoofing to spread election-related misinformation.

“Political consultant Steve Kramer was responsible for the calls and now faces a $6 million proposed fine for perpetrating this illegal robocall campaign on January 21, 2024,” the FCC said in a statement.

The FCC alleged that Kramer orchestrated a robocall campaign that featured a deepfake of President Joe Biden’s voice, urged New Hampshire voters not to participate in the January primary, asking them to “save your vote for the November election.”

Kramer’s action, conducted just two days before the presidential primary, violated the “Truth in Caller ID Act,” the FCC said. This law prohibits the transmission of false or misleading caller ID information with the intent to defraud, cause harm, or wrongfully obtain value.

 “We will act swiftly and decisively to ensure that bad actors cannot use U.S. telecommunications networks to facilitate the misuse of generative AI technology to interfere with elections, defraud consumers, or compromise sensitive data,” Loyaan A Egal, chief of the Enforcement Bureau and chair of the Privacy and Data Protection Task Force at FCC said in the statement.

The FCC is also taking action against Lingo Telecom for its role in facilitating the illegal robocalls, the statement added.

“Lingo Telecom transmitted these calls, incorrectly labeling them with the highest level of caller ID attestation, making it less likely that other providers could detect the calls as potentially spoofed.  The Commission brought a separate enforcement action today against Lingo Telecom for apparent violations of STIR/SHAKEN for failing to utilize reasonable “Know Your Customer” protocols to verify caller ID information in connection with Mr. Kramer’s illegal robocalls.”

The Commission has made clear that calls made with AI-generated voices are “artificial” under the Telephone Consumer Protection Act (TCPA), confirming that the FCC and state Attorneys General have the needed tools to go after bad actors behind these nefarious robocalls, the statement added. “In addition, the FCC launched a formal proceeding to gather information on the current state of AI use in calling and texting and ask questions about new threats, like robocalls.”

Echoes of a wider debate

This incident reignites concerns over the potential misuse of deepfakes, a technology that can create realistic and often undetectable audio and video forgeries.

Earlier this month, actress Scarlett Johansson raised similar concerns alleging OpenAI using her voice without consent in its AI application. She had alleged that the voice behind “Sky” voice chat sounded “eerily similar” to her. However, OpenAI quickly refuted this allegation.

“The voice of Sky is not Scarlett Johansson’s, and it was never intended to resemble hers,” OpenAI CEO Sam Altman said in a statement. “We cast the voice actor behind Sky’s voice before any outreach to Ms. Johansson. Out of respect for Ms. Johansson, we have paused using Sky’s voice in our products.”

Meanwhile, the ChatGPT maker had paused the voice of “Sky,” the statement added.

“Johansson’s case highlights broader ethical and legal challenges surrounding AI-generated content and the need for stringent regulations to protect individuals’ privacy and identities,” said Faisal Kawoosa, founder and chief analyst at the technology research and consulting firm, Techarc.

The IBM mainframe is the latest platform to gain a generative AI-based assistant to simplify automation and help customers with routine tasks. IBM rolled out its new watsonx Assistant for Z v1, which will tie together automation coding and watsonx AI technology to ease mainframe operations.

“Users can seamlessly import existing trusted automation created with Ansible, job control language (JCL), and REXX as skills into the product catalog,” wrote Parul Mishra, vice president, product management with IBM Digital Labor, and Skyla Loomis, vice president, IBM Z Software, in a blog about the new assistant.

“These skills can be invoked via a contextually aware chat experience, using watsonx Orchestrate’s extensible automation framework that improves productivity and accelerates efficiency on the mainframe,” they wrote. WatsonX lets customers build and manage AI resources and workflows.

watsonx Assistant for Z uses a chat-focused granite 13b.labrador generative AI Large Language Foundation model. This model is an enterprise-level, English-language system that is trained with a dataset of more than 2.5 trillion tokens, including IBM Z-specific content. It uses Z domain-specific retrieval-augmented generation (RAG) to provide curated content to users, according to Mishra and Loomis. For example, the assistant can generate answers to IBM Z users’ questions in Z-specific content as well as recommend relevant automated skills that can be acted upon during the conversation.

An AI assistant builder offers access to a low-code generative AI and automation studio to stand-up large language model (LLM) powered assistants that are grounded in business context, data, and automation, the bloggers stated.

“The AI assistant builder enables you to extend your investments in automation and scale adoption through a highly conversational interface that is infused with AI that you can trust. Generative AI capabilities are delivered natively by IBM’s trusted and transparent LLMs fine-tuned for specific business functions,” the bloggers stated. “Builders have the flexibility and control to decide when and how LLMs are used via content grounding, AI-guided actions, and conversational skills.”

Going forward, IBM said it intends to develop an IBM watsonx Assistant for Z that runs on IBM Linux on Z. The ideas are the same for the Linux distribution in that customers can simplify operations and development on IBM Linux on Z.

IBM Z is just the latest in myriad AI-based assistants to be offered from a variety of vendors including Cisco, Juniper, Extreme, Microsoft, Fortinet and many more.

A recent Gartner report stated that by 2028, 75% of enterprise software engineers will use AI code assistants, up from less than 10% in early 2023. Sixty-three percent of organizations are currently piloting, deploying or have already deployed AI code assistants.

While some enterprises looking at private 5G have followed 5G’s Open RAN development as part of their assessment or deployment plans, most tell me the technology is for network operators, and they pay it little mind. The announcement that IBM is buying HashiCorp stirred a lot more enterprise interest. As it turns out, both are linked to issues with the very concept of open technology that most enterprises depend upon.

Why do enterprises say they like “open” technology? Of 422 technology professionals who offered me an opinion in the last year, 403 said it was to “prevent lock-in” or “keep vendors honest.” I took multiple responses to the question, and another 385 said that it was for “innovation,” and 247 said “lower costs.” Through all the responses, there was a strong thread that vendors can’t be trusted. They will gouge you on pricing, stall innovation to boost their profits, pull products, slip schedules…you get the picture. Who’s afraid of the big bad vendor wolf? Pretty much every enterprise.

OK, but how do you square all of this “vendors are the axis of evil” stuff with this point: 379 of that same group, in the same period, said that over the last several years, they’d purchased a larger percentage of their technology from their dominant vendor? I didn’t get a lot of spontaneous guidance on that question from enterprises, but of 54 network operators I questioned on Open RAN, I did get some interesting views, and I think their input could likely apply to enterprises as well.

Of the 54 operators, 50 said that multi-vendor integration was getting steadily more difficult and costly, and that open RAN solutions were explicitly multi-vendor. Even more, 52 of 54, said that the “innovation” they’d expected from Open RAN hadn’t materialized (22 said) or was “significantly less” than expected (30 said). In all, 47 of the 54 operators admitted their policies were moving them toward a smaller number of vendors in 5G RAN, and 29 of these admitted they’d likely reduce the field to one vendor.

What about that greedy, lazy, exploiting, vendor you have to protect against? One CFO put it best: “We found vaccinating against vendor exploitation was too expensive, so we decided to wait till we caught the disease and then treat it.” This meant that open technology was a promise of an exit path from a bad situation, to be taken only if needed. Wait until your vendor does something you don’t like, then threaten them with migration to another member of an open ecosystem. Beat them up with openness, but go elsewhere only if you absolutely have to, and don’t try to pursue open-model tech on your own. Pick a giant source you can threaten.

OK, that’s the network operator space. Do enterprises really think the same way? For that, I offer IBM’s experience, first with Red Hat and now with HashiCorp. Why would a smart vendor buy a company whose product(s) are open source? The only possible answer is that the buyers of the product would rather get it in commercial form than in open form. Part of the reason is that it’s hard for enterprises to hire and retain specialists on all the open-source tools they might want. Part is because they want a deep-pocketed, legally-liable backer for their critical tools. Self-insurance, or being your own lawyer, is always a risk.

Some open-source supporters have been upset about license changes and other shifts that contaminate the original value proposition, but the biggest contributor to that pollution may be buyer/user behavior. Among large enterprises, in-house open-source expertise used to be subjectively rated as “high” by every player just a couple decades ago, but today less than one-third offer that kind of optimism. And among SMBs, the number who think they have and can retain the skilled people to manage their own open-source tools is in the statistical noise level.

To use open source, more and more enterprises rely on some large organization to back their bid, and that puts pressure on the open-source initiatives to adopt things like dual licensing to permit charging for some version of the tools. Enterprises cite the problems with openness as a reason why white-box network devices haven’t swept the market. Who backs a mixture of chips, generic hardware, and a couple layers of open-source software?

And how do you innovate in a complex ecosystem? Could a team have painted the Mona Lisa? I asked an enterprise CIO that question, and got a question in return, a better one. “Who would go to see it?” Putting that point another way, not only does the open-model innovation theory have to ask how a vision of a system can emerge from a series of parts suppliers, it also has to ask who has the incentive to promote it, and whose name can give it credibility.

That metaphor suggests another question, though: “How would the team know what they were painting?”

I’ve seen a lot of open-technology projects that were founded to do something useless, impossible, or both. Even more that had a potentially worthy notion that demanded organized initiatives to validate and promote it, which never came to be. Open RAN may well be a glaring example. OK, so we open up parts of the RAN model that the 3GPP work kept closed. What actual benefit, if we define “benefit” as something that can make a business case, does that create? Most suggestions fall into the “glittering generality” category, and the few more concrete ones have failed to pan out, so you have to question them all. Yes, it created a play for smaller vendors, but what kind of vendor is winning the 5G deals? Ones whose headcounts could fill a small city.

We have to go back to the opening point, which is that buyers see open-model technology as the final defense against vendor misbehavior. It’s a great theory, but what good is a defense alliance with a player who you allow to starve to death while you’re waiting for an attack? And how many guarantors of your defense will emerge if that’s the fate that awaits them all?

You can’t blame buyers here. There has to be a real value, not a simple threat value, to open-model technology. It’s up to those who want to advance it to ensure it can deliver that real value, which means that the promise of innovation has to be fulfilled. That means open projects, including open-source software, has to obey four rules:

• First, define a project around a need that’s truly and immediately valuable.
• Second, get everyone’s agreement to promote the business case, not the technology.
• Third, aim every project step at realizing that business case.
• Fourth, establish how collective support credibility can be established and maintained.

As a provider/promoter of open technology, you need to ensure these rules are followed. And you need to decide up front how media and web outreach will be managed so the project speaks with one voice.

As a consumer of that technology, you need to look at the things you’re thinking of adopting and be sure they address all these points – to ensure their success and your own.

Gaining firm control over a multicloud network environment can deliver performance, efficiency, and security gains – but it won’t be easy.

Managing a multicloud network is a highly demanding task, says Robert Orshaw, a managing director at Deloitte Consulting. “The primary challenges include maintaining a holistic view of security controls, meeting regulatory standards across varied cloud platforms, securing data across multiple providers to prevent unauthorized access or accidental loss, and managing the architectural and operational complexity of utilizing multiple providers,” he explains. “Additionally, a multicloud environment can introduce cost inefficiencies due to resource sprawl across different platforms.”

Coping with complexity

Several factors contribute to making multicloud network management a complex and, at times, frustrating task. “These include skills and expertise gaps, complex tools and application integrations, security challenges, vulnerabilities, and infrastructure visibility across networks and beyond,” says Venkata Achanti, vice president, cloud, and custom applications, at business advisory firm Capgemini Americas. “Additionally, payment processes can differ for each cloud provider, making it difficult to manage costs.”

Multiple vendors are leading the charge of managing and/or running multicloud environments, including Cisco, Datadog, HPE, Nutanix, Palo Alto Networks, Red Hat, and VMware.

“These players have specific roles at various tiers of multicloud, from log to traffic governance,” says Mohammad Wasim, global practice lead for cloud, infrastructure, and security with digital transformation consulting firm Publicis Sapient. Many of these vendors have mature feature sets. “The ones investing to bring disruption are Nvidia and AWS on their silicon chips, which will fundamentally change the [multicloud] performance level as well as enable moving the processing of data to the edge.”

Multicloud network management poses significant challenges, primarily due to the inherent complexity and scale of handling diverse cloud environments, Wasim observes. “Organizations adopting multicloud strategies often grapple with managing different technologies, architectures, and vendors across various cloud providers,” he explains. “This complexity introduces challenges in orchestrating seamless interactions and ensuring compatibility.”

While each cloud provider does its best to make networking simple across clouds, all have very nuanced differences and varied best practices for approaching the same problem, says Ed Wood, global enterprise network lead at business advisory firm Accenture. This makes being able to create enterprise-ready, secured networks across the cloud challenging, he adds.

Wasim believes that a lack of intelligent data utilization at crucial stages, from data ingestion to proactive management, further complicates the process. “The sheer scale of managing resources, coupled with the dynamic nature of cloud environments, makes it challenging to achieve optimal performance and efficiency.”

Making network management even more challenging is a lack of clarity on roles and responsibilities. This can be attributed to an absence of agreement on shared responsibility models, Wasim says. As a result, stakeholders, including customers, cloud service providers, and any involved third parties, might each hold different views on responsibility and accountability regarding data compliance, controls, and cloud operations management.

“Furthermore, with multicloud, organizations are seeing a convergence in roles and job descriptions,” he notes. For example, a network engineer may be required to oversee the entire cloud infrastructure instead of just a single localized network. “If an organization doesn’t have the right talents and skillsets, managing the network in a multicloud environment may become challenging.”

Frustrations and blind spots

Effective multicloud network management can also be derailed by mistakes made by team members.

“With today’s ease of access to cloud platforms, virtually anyone in an organization can create significant governance blind spots, often leading to shadow IT-type risks,” warns Orshaw. “With virtually anyone able to deploy cloud elements, the effort required for adequate governance becomes a major frustration, emphasizing the vital need for a comprehensive, organization-wide multicloud strategy.” Such a strategy forces IT to deploy an efficient and dependable multicloud network management platform. “The difficulty is in finding platforms that have similar strengths across all cloud providers.”

Wasim notes that multicloud management often includes operational challenges related to migration, a lack of clarity on roles and responsibilities, technical complexities arising from managing multiple cloud architectures, and cloud costs.

“A multicloud environment, by nature, is decentralized, and having a cloud strategy on paper is not enough.” He observes that blind spots often emerge due to a “lack of cadence” between procurement managers, the cloud team, the IT team, CFO team members, and the cloud service provider.

Network issues are another concern. A typical cloud infrastructure’s scale can make it difficult for the network team to have full visibility into the data flow, Wasim says. Therefore, both infrastructure and operations leaders should be investing in AIOps platforms to improve network visibility.

Developing and advancing the skillsets of existing talent to manage multicloud can be time consuming and costly. “Leaders must increase their focus on acquiring, training, and retaining such talent, which can be arduous,” Achanti says.

Essential tools

When deploying a multicloud network management platform, Orshaw advises using strong tools to detect, monitor, and manage all essential assets. He also advocates a robust cybersecurity strategy that’s integrated directly into the network platform.

To improve both multicloud network management and security, Achanti says it’s advisable to use dashboards and other tools  that can provide thorough visibility into infrastructure and application assets. He also recommends deploying processes that target cybersecurity and compliance requirements. “In a multicloud environment, more effort is required when developing procedures that reinforce security, because each cloud vendor has unique techniques native to their cloud environment,” Achanti says.

Critical features

A multicloud network management tool should provide integrated visibility and management capabilities delivered as a cohesive layer above each cloud provider’s native tooling, Orshaw says. “These essential single-pane-of-glass solutions allow organizations to utilize the powerful tools that the hyperscalers offer natively.”

When selecting a management tool, Wood suggests looking for event management and correlation, AI insights, external sources ingestion, and strong visualization capabilities. “It’s key to be able to integrate with various third-party vendor solutions and orchestrate them along cloud service provider native services,” he says. “While cloud service providers aim to provide, and continuously improve. cloud native services, almost every large enterprise will have third-party provider needs.”

Orshaw recommends adopting a disciplined FinOps practice that not only provides the right amount of governance, but also helps optimize spend across cloud service providers. “The main characteristics of a multicloud network management platform should include unified management, automation, standardization, visibility and monitoring, security, vendor management, training and skills development, and continuous optimization.”

Perhaps the most important decision in any multicloud strategy is handling the orchestration between various cloud technologies. “It’s not about how a single pane shows the dashboards of multiple and different clouds, but how we engineer the solution so that it’s set up in a uniform and standard manner,” Wasim says. Prioritizing orchestration not only improves multicloud management, but it also reduces the business impact when downtime incidents occur, since the approach makes it easier to troubleshoot issues. Meanwhile, automation tools are necessary to streamline workflows and ensure seamless collaboration across diverse cloud environments.

Current trends

AI is the most important multicloud network management trend, Orshaw says. “Using AI technologies ultimately leads to detection and self-healing without any human intervention.”

Wasim agrees. “An ideal multicloud management solution should incorporate AI-based features that excel in anomaly detection, event management, and correlation in implementing use cases with different technology frameworks.” Such capabilities, he notes, enable multicloud adopters to proactively identify and address issues at stages as early as the design phase, minimizing build-out effort and downtime while improving overall performance. What’s most critical, Wasim says, is that strong AI-supported data security measures are implemented consistently and are designed to provide effective governance capabilities to safeguard against potential risks and ensure compliance with data protection regulations.

Traditional monitoring tools may not be enough to allow multicloud adopters to fully understand data flow, as well as any vulnerabilities attributable to configuration integration or mismanagement, Wasim warns. “AI-powered tools help in understanding patterns in large datasets while automating remediation and, therefore, improve key metrics, such as time-to-resolve network issues,” he says. “AI, along with ML, is now helping the IT footprint with proactive maintenance.” These technologies can also project business outcomes or events.

Besides detecting anomalies, AI-based solutions can also contribute to intelligent decision-making, allowing multicloud adopters to navigate the complexities of their environments with enhanced agility and precision, Wasim says. “What’s now being tested is generative AI, which is lowering the expectation of tech capability in CloudOps engineering,” he notes. “As efficiencies in tools such as CodeWhisperer, Copilot, or Amazon Q improve, we will see exponential efficiency gains.”

“Selecting the right vendor to accelerate your multicloud journey comes down to identifying the organizations that have a solid track record of excellence and expertise in the sector and capability areas most relevant to your organization,” Orshaw says.

The reliability of services delivered by ISPs, cloud providers and conferencing services is critical for enterprise organizations. ThousandEyes, a Cisco company, monitors how providers are handling any performance challenges and provides Network World with a weekly roundup of events that impact service delivery. Read on to see the latest analysis, and stop back next week for another update.

(Note: We have archived prior-year updates, including the 2023 outage report and our coverage during the Covid-19 years, when we began tracking the performance of cloud providers and ISPs.)

Internet report for May 13-19, 2024

ThousandEyes reported 227 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of May 13-19. That’s up a sizable 43% from 159 outages the week prior. Specific to the U.S., there were 84 outages, which is up 27% from 66 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages increased from 105 to 157 outages, a 50% leap compared to the week prior. In the U.S., the number of ISP outages increased 43% from 40 to 57 outages.

Public cloud network outages: Globally, cloud provider network outages increased from 18 to 33 outages. In the U.S., cloud provider network outages increased from nine to 10.

Collaboration app network outages: Globally, collaboration app network outages decreased from seven to six outages. In the U.S., collaboration app network outages increased from two to three outages.

Two notable outages

On May 19, Cogent Communications, a U.S.-based multinational transit provider, experienced an outage that impacted multiple downstream providers as well as Cogent customers across various regions, including the U.S., the U.K., Mexico, Spain, Singapore, Brazil, Turkey, Germany, Switzerland, the Netherlands, China, Portugal, Chile, Peru, Canada, India, Thailand, South Korea, France, and Japan. The outage, which lasted 19 minutes, was first observed around 4:55 PM EDT and initially centered on Cogent nodes located in Nashville, TN. Five minutes after first being observed, the nodes experiencing outage conditions expanded to include nodes in Houston, TX, Atlanta, GA, New York, NY, Washington, D.C., and Philadelphia, PA. Five minutes later, the number of locations with nodes exhibiting outage conditions expanded again, this time including nodes in Bilbao, Spain, Dallas, TX, Phoenix, AZ, Charlotte, NC, and Indianapolis, IN. As a result, the number of impacted customers and providers increased. The outage was resolved around 5:15 PM EDT. Click here for an interactive view.

On May 16, Rackspace Technology, a U.S. managed cloud computing provider headquartered in San Antonio, Texas, experienced a series of outages over a period of four hours and fifteen minutes that impacted multiple downstream providers, as well as Rackspace customers within multiple regions including the U.S., Canada, Mexico, Germany, Singapore, Chile, Brazil, France, Spain, South Africa, Vietnam, Turkey, and Switzerland. The outage, lasting a total of 28 minutes, was first observed around 5:00 AM EDT and appeared to center on Rackspace nodes located in Chicago, IL. The outage was cleared around 9:15 AM EDT. Click here for an interactive view.

Internet report for May 6-12, 2024

ThousandEyes reported 159 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of May 6-12. That’s down 5% from 151 outages the week prior. Specific to the U.S., there were 66 outages, which is up 32% from 50 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages declined from 113 to 105 outages, a 7% decrease compared to the week prior. In the U.S., the number of ISP outages increased slightly (3%) from 39 to 40 outages.

Public cloud network outages: Globally, cloud provider network outages increased from 15 to 18 outages. In the U.S., cloud provider network outages jumped from two to nine.

Collaboration app network outages: Globally, collaboration app network outages decreased from nine to seven outages. In the U.S., collaboration app network outages fell from three to two outages.

Two notable outages

On May 8, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across the U.S. The outage was first observed at around 7:45 PM EDT and appeared to center on Time Warner Cable nodes located in New York, NY.  Five minutes after first being observed, the number of nodes located in New York, NY, exhibiting outage conditions increased. The outage lasted a total of 23 minutes and was cleared at around 8:10 PM EDT. Click here for an interactive view.

On May 7, Comcast Communications experienced an outage that impacted a number of downstream partners and customers across the U.S. and Canada. The outage, lasting 8 minutes, was first observed around 1:20 AM EDT and appeared to be centered on Comcast nodes located in Chicago, IL and Ashburn, VA. Five minutes into the outage, the nodes located in Ashburn, VA, appeared to clear, leaving the nodes located in Chicago, IL, exhibiting outage conditions. The apparent decrease of nodes exhibiting outage conditions appeared to coincide with a decrease in the number of impacted downstream customers and partners. The outage was cleared around 1:30 AM EDT. Click here for an interactive view.

Internet report for April 29- May 5, 2024

ThousandEyes reported 151 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of April 29- May 5. That’s down slightly (3%) from 156 outages the week prior. Specific to the U.S., there were 50 outages, which is down 7% from 54 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages increased from 104 to 113 outages, a 9% increase compared to the week prior. In the U.S., the number of ISP outages increased 5% from 37 to 39 outages.

Public cloud network outages: Globally, cloud provider network outages decreased from 22 to 15 outages. In the U.S., cloud provider network outages decreased from six to two.

Collaboration app network outages: Globally, collaboration app network outages increased from eight to nine outages. In the U.S., collaboration app network outages fell from four to three outages.

Two notable outages

On April 29, NTT America, a global Tier 1 ISP and subsidiary of NTT Global, experienced an outage that impacted some of its customers and downstream partners in multiple regions, including the U.S., Japan, South Korea, China, Taiwan, Singapore, the Netherlands, Hungary, Turkey, Brazil, India, Argentina, Australia, the U.K., Thailand, Malaysia, Mexico, and Canada. The outage, lasting 24 minutes, was first observed around 2:40 PM EDT and appeared to initially center on NTT nodes located in San Jose, CA. Around five minutes into the outage, the nodes located in San Jose, CA, appeared to clear, and were replaced by nodes located in Tokyo, Japan, in exhibiting outage conditions. Ten minutes after first being observed, the nodes located in Tokyo, Japan, were joined by nodes located in Osaka, Japan, Singapore, Dallas, TX, and Los Angeles, CA, in exhibiting outage conditions. The outage was cleared around 3:05 PM EDT. Click here for an interactive view.

On April 29, Cogent Communications, a multinational transit provider based in the US, experienced an outage that impacted multiple downstream providers and customers across various regions, including the U.S., Brazil, the U.K., Canada, Chile, Mexico, Japan, Germany, Spain, and France. The outage, lasting for a total of one hour and 12 minutes, was divided into two occurrences over a period of 35 minutes. The first occurrence was observed around 2:45 AM EDT and initially seemed to be centered on Cogent nodes located in Ashburn, VA, and Washington, D.C. Five minutes into the outage, the nodes located in Ashburn, VA, appeared to clear and were replaced by nodes located in Baltimore, MD, New York, NY, and Phoenix, AZ, along with nodes located in Washington, D.C., in exhibiting outage conditions. This increase in nodes exhibiting outage conditions also appeared to coincide with an increase in the number of downstream customers, partners, and regions impacted. Twenty minutes after appearing to clear, nodes located in New York, NY, and Washington, D.C., were joined by nodes located in Houston, TX, in exhibiting outage conditions. The outage was cleared around 3:20 AM EDT. Click here for an interactive view.

Internet report for April 22-28, 2024

ThousandEyes reported 156 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of April 22-28. That’s down 8% from 170 outages the week prior. Specific to the U.S., there were 54 outages, which is down 36% from 85 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages increased from 99 to 104 outages, a 5% increase compared to the week prior. In the U.S., the number of ISP outages decreased 31% from 54 to 37 outages.

Public cloud network outages: Globally, cloud provider network outages remained the same as the week prior, recording 22 outages. In the U.S., cloud provider network outages decreased from 10 to six.

Collaboration app network outages: Globally, collaboration app network outages decreased from nine to eight outages. In the U.S., collaboration app network outages stayed at the same level as the week before: four outages.

Two notable outages

On April 26, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across the U.S. The outage was distributed across two occurrences over a twenty-five-minute period and was first observed at around 7:45 PM EDT and appeared to center on Time Warner Cable nodes located in New York, NY.  Ten minutes after first being observed, the number of nodes located in New York, NY, exhibiting outage conditions increased. The outage lasted a total of 17 minutes and was cleared at around 8:10 PM EDT. Click here for an interactive view.

On April 24, NTT America, a global Tier 1 ISP and subsidiary of NTT Global, experienced an outage that impacted some of its customers and downstream partners in multiple regions, including the U.S., Germany, India, China, Hong Kong, Canada, and Japan. The outage, lasting 9 minutes, was first observed around 7:15 AM EDT and appeared to initially center on NTT nodes located in San Jose, CA. Around five minutes into the outage, the nodes located in San Jose, CA, were joined by nodes located in Dallas, TX, in exhibiting outage conditions. The outage was cleared around 7:25 AM EDT. Click here for an interactive view.

Internet report for April 15-21, 2024

ThousandEyes reported 170 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of April 15-21. That’s up 11% from 161 outages the week prior. Specific to the U.S., there were 85 outages, which is up 18% from 72 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages decreased from 107 to 99 outages, a 7% decline compared to the week prior. In the U.S., the number of ISP outages climbed 6% from 51 to 54 outages.

Public cloud network outages: Both globally (22) and in the U.S. (10), cloud provider network outages remained the same as the week prior.

Collaboration app network outages: Globally, collaboration app network outages increased from five to nine outages. In the U.S., collaboration app network outages increased from two to four outages.

Two notable outages

On April 20, Cogent Communications, a multinational transit provider based in the US, experienced an outage that impacted multiple downstream providers and its own customers across various regions, including the US, Canada, Portugal, Germany, the Netherlands, Luxembourg, South Africa, Hong Kong, Singapore, the U.K., Italy, France, and Spain. The outage, lasting a total of one hour and 32 minutes, was divided into a series of occurrences over a period of two hours and 28 minutes. The first occurrence was observed around 10:55 PM EDT and initially seemed to be centered on Cogent nodes located in Seattle, WA, Portland, OR, and Hong Kong. Five minutes into the outage, the nodes located in Hong Kong appeared to clear and were replaced by nodes located in Minneapolis, MN, and Cleveland OH, in exhibiting outage conditions. Thirty-five minutes into the first occurrence, the number of nodes exhibiting outage conditions increased to include nodes located in Seattle, WA, Washington, D.C., Minneapolis, MN, Cleveland, OH, Boston, MA, and Bilbao, Spain. This increase in nodes exhibiting outage conditions also appeared to coincide with an increase in the number of downstream customers, partners, and regions impacted. A second occurrence was observed around five minutes after the issue initially appeared to have cleared. This second occurrence lasted approximately fourteen minutes and seemed to initially be centered around nodes located in Cleveland, OH. Around five minutes into the second occurrence, nodes located in Cleveland, OH, appeared to be temporarily replaced by nodes located in Seattle, WA, and Chicago, IL, before they themselves were replaced once again by nodes located in Cleveland, OH.  Around 15 minutes after appearing to clear, a third occurrence was observed, this time appearing to be centered around nodes located in Bilbao, Spain, and Cleveland, OH. The outage was cleared around 1:25 AM EDT. Click here for an interactive view.

On April 17, NTT America, a global Tier 1 ISP and subsidiary of NTT Global, experienced an outage that impacted some of its customers and downstream partners across multiple regions including the U.S., the U.K., the Netherlands, and Germany. The outage, lasting 17 minutes, was first observed around 2:55 AM EDT and appeared to initially center on NTT nodes located in Seattle, WA. Five minutes into the outage nodes located in Seattle, WA, were joined by nodes located in Dallas, TX, in exhibiting outage conditions. The outage was cleared around 3:15 AM EDT. Click here for an interactive view.

Internet report for April 8-14, 2024

ThousandEyes reported 161 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of April 8-14. That’s up 11% from 154 outages the week prior. Specific to the U.S., there were 72 outages, which is up 4% from 69 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages increased from 97 to 107 outages, a 10% increase compared to the week prior. In the U.S., the number of ISP outages climbed 13% from 45 to 51 outages.

Public cloud network outages: Globally, cloud provider network outages increased from 16 to 22 outages. In the U.S., cloud provider network outages decreased from 14 to 10 outages.

Collaboration app network outages: Globally, collaboration app network outages remained at the same level as the week prior, recording 5 outages. In the U.S., collaboration app network outages decreased from three to two outages.

Two notable outages

On April 8, Rackspace Technology, a U.S. managed cloud computing provider headquartered in San Antonio, Texas, experienced an outage that impacted multiple downstream providers, as well as Rackspace customers within multiple regions including the U.S., Japan, Vietnam, Spain, Canada, Germany, Singapore, France, the Netherlands, the U.K., Brazil, and South Africa. The outage, lasting a total of 14 minutes, was first observed around 9:00 AM EDT and appeared to center on Rackspace nodes located in Chicago, IL. Around five minutes into the outage, the number of nodes located in Chicago, IL, exhibiting outage conditions, appeared to clear. This decrease in impacted nodes appeared to coincide with a reduction of impacted regions. The outage was cleared around 9:15 AM EDT. Click here for an interactive view.

On Apr 10, GTT Communications, a Tier 1 ISP headquartered in Tysons, VA, experienced an outage that impacted some of its partners and customers across multiple regions, including the U.S., the U.K., Brazil, and Canada. The outage, lasting 9 minutes, was first observed around 8:10 AM EDT and appeared to initially be centered on GTT nodes located in Los Angeles, CA. Around five minutes into the outage the nodes located in Los Angeles appeared to clear and were replaced by nodes located in New York, NY, in exhibiting outage conditions. This change in location of nodes exhibiting outage conditions appeared to coincide with an increase in the number of impacted regions, downstream partners and customers. The outage was cleared around 8:20 AM EDT. Click here for an interactive view.

Internet report for April 1-7, 2024

ThousandEyes reported 145 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of April 1-7. That’s up 23% from 118 outages the week prior. Specific to the U.S., there were 69 outages, which is up 21% from 57 outages the week prior. Here’s a breakdown by category:

ISP outages: Both globally and in the U.S., ISP outages increased by 45% compared to the week prior. Globally, the number of ISP outages climbed from 67 to 97. In the U.S., the number of ISP outages jumped from 31 to 45 outages.

Public cloud network outages: Globally, cloud provider network outages declined slightly from 17 to 16 outages. In the U.S., they remained at the same level (14) as the previous week.

Collaboration app network outages: Globally, collaboration app network outages fell from 13 to five outages. In the U.S., collaboration app network outages dropped from eight to three outages.

Two notable outages

On April 2, Hurricane Electric, a network transit provider headquartered in Fremont, CA, experienced an outage that impacted customers and downstream partners across multiple regions, including the U.S., Taiwan, Australia, Germany, Japan, the U.K., Ireland, India, and China. The outage, lasting 12 minutes, was first observed around 12:40 PM EDT and initially appeared to center on Hurricane Electric nodes located in New York, NY, Los Angeles, CA, and San Jose, CA. Five minutes into the outage, the nodes exhibiting outage conditions expanded to include nodes located in Chicago, IL, and Ashburn, VA. This coincided with an increase in the number of downstream partners and countries impacted. The outage was cleared around 12:55 PM EDT. Click here for an interactive view.

On April 2, BT, a multinational Tier 1 ISP headquartered in London, U.K., experienced an outage on their European backbone that impacted customers and downstream partners across multiple regions, including the U.S., the U.K., Switzerland, Spain, and Germany. The disruption, lasting 24 minutes, was first observed around 7:20 PM EDT and appeared to center on nodes located in London, England. Click here for an interactive view.

Internet report for March 25-31, 2024

ThousandEyes reported 118 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of March 18-24. That’s down 28% from 164 outages the week prior. Specific to the U.S., there were 57 outages, which is down slightly from 58 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages was nearly cut in half, falling from 128 to 67 outages, a 48% decrease compared to the week prior. In the U.S., the number of ISP outages fell slightly from 43 to 31 outages, a decline of 28%.

Public cloud network outages: Globally, total cloud provider network outages nearly tripled, jumping from six to 17 outages. In the U.S., cloud provider network outages jumped from three to 14 outages

Collaboration app network outages: Globally, collaboration app network outages more than doubled, increasing from six to 13. Similarly, in the U.S., collaboration app network outages doubled from four to eight.

Two notable outages

On March 29, Arelion (formerly known as Telia Carrier), a global Tier 1 ISP headquartered in Stockholm, Sweden, experienced an outage that impacted customers and downstream partners across multiple regions, including the U.S., the Netherlands, and Japan. The disruption, lasting a total of 8 minutes, was first observed around 5:26 AM EDT and appeared to center on nodes located in Phoenix, AZ. The outage was cleared around 5:35 AM EDT. Click here for an interactive view.

On March 29, Cogent Communications, a U.S. based multinational transit provider, experienced an outage that impacted multiple downstream providers as well as Cogent customers across multiple regions, including the U.S., Canada, Germany, and Japan. The outage, lasting 9 minutes, was first observed around 1:45 AM EDT and appeared to initially center on Cogent nodes located in San Francisco, CA, Salt Lake City, UT, and Seattle, WA. Five minutes after first being observed, the nodes located in San Francisco, CA, Salt Lake City, UT and Seattle, WA, appeared to recover and were replaced by nodes located in Kansas City, MO in exhibiting outage conditions. As a result, the number of customers and providers impacted was reduced. The outage was cleared around 1:55 AM EDT. Click here for an interactive view.

Internet report for March 18-24, 2024

After a spike the week before, global outages decreased last week. ThousandEyes reported 164 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of March 18-24. That’s down 20% from 206 outages the week prior. Specific to the U.S., there were 58 outages, which is down 33% from 87 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages fell slightly from 131 to 128 outages, a 2% decrease compared to the week prior. In the U.S., the number of ISP outages fell slightly from 46 to 43 outages.

Public cloud network outages: Globally, cloud provider network outages decreased from 10 to six outages. In the U.S., they decreased from six to three outages.

Collaboration app network outages: Globally, collaboration app network outages fell dramatically from 34 to six outages. In the U.S., collaboration app network outages dropped from 28 to four outages.

Two notable outages

On March 20, Cogent Communications, a multinational transit provider based in the US, experienced an outage that impacted multiple downstream providers and its own customers across various regions, including the US, Italy, Saudi Arabia, France, Germany, Canada, Hong Kong, Luxembourg, Chile, Brazil, Kenya, Singapore, Mexico, Switzerland, Spain, Australia, Finland, Japan, Ireland, and Norway. The outage occurred for a total of 24 minutes, divided into three occurrences over a period of one hour and fifteen minutes. The first occurrence of the outage was observed around 12:50 AM EDT and initially seemed to be centered on Cogent nodes located in Frankfurt, Munich, and Hamburg in Germany, in Paris, France, and Kyiv, Ukraine. A second occurrence was observed around fifteen minutes after the issue initially appeared to have cleared. This second occurrence lasted approximately eighteen minutes and seemed to be centered around nodes located in Frankfurt, Munich, and Hamburg, Germany. Around ten minutes into the second occurrence, nodes located in Frankfurt, Munich and Hamburg, Germany, were joined by nodes located in Nuremberg, Germany, San Francisco, CA, San Jose, CA, Zurich, Switzerland, Amsterdam, the Netherlands, and Paris, France in exhibiting outage conditions. Around 30 minutes after appearing to clear, a third occurrence was observed, this time appearing to be centered around nodes located in Toronto, Canada. The outage was cleared around 2:05 AM EDT. Click here for an interactive view.

On March 24, Hurricane Electric, a network transit provider headquartered in Fremont, CA, experienced an outage that impacted customers and downstream partners across multiple regions, including the U.S., China, Australia, Germany, the U.K., and Japan. The outage, first observed around 1:10 PM EDT, lasted 7 minutes in total and appeared to center on Hurricane Electric nodes located in New York, NY, and San Jose, CA. The outage was cleared at around 1:20 PM EDT. Click here for an interactive view.

Internet report for March 11-17, 2024

After weeks of decreasing, global outages increased significantly last week. ThousandEyes reported 206 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of March 11-17. That’s up 45% from 142 outages the week prior. Specific to the U.S., there were 87 outages, which is up 38% from 63 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages increased from 91 to 131 outages, a 44% increase compared to the week prior. In the U.S., the number of ISP outages climbed slightly from 44 to 46 outages.

Public cloud network outages: Globally, cloud provider network outages increased from six to 10 outages. In the U.S., they increased from four to six outages.

Collaboration app network outages: Globally, collaboration app network outages spiked from six to 34 outages. In the U.S., collaboration app network outages jumped from 3 to 28 outages.

Two notable outages

On March 16, Cogent Communications, a U.S. based multinational transit provider, experienced an outage that impacted multiple downstream providers as well as Cogent customers across multiple regions, including the U.S., Ireland, the U.K., Sweden, Austria, Germany, and Italy. The outage, lasting a total of 12 minutes, was divided into two occurrences over a one-hour and ten-minute period. The first occurrence was observed at around 6:30 PM EDT and appeared to initially be centered on Cogent nodes located in Baltimore, MD and New York, NY. Five minutes into the first occurrence, the nodes located in New York, NY, were replaced by nodes located in Philadelphia, PA, in exhibiting outage conditions. One hour after the issue initially appeared to have cleared, a second occurrence was observed. This second occurrence lasted approximately four minutes and appeared to be centered around nodes located in Baltimore, MD, Philadelphia, PA, New York, NY, and Newark, NJ. The outage was cleared around 7:45 PM EDT. Click here for an interactive view.

On March 12, Hurricane Electric, a network transit provider headquartered in Fremont, CA, experienced an outage that impacted customers and downstream partners across the U.S. and Canada. The outage, first observed around 2:00 AM EDT, lasted 7 minutes in total and was divided into two occurrences over a thirty-minute period. The first occurrence appeared to initially center on Hurricane Electric nodes located in Chicago, IL. Twenty minutes after appearing to clear, the nodes located in Chicago, IL, were joined by nodes located in Seattle, WA in exhibiting outage conditions. This increase in impacted nodes appeared to coincide with an increase in the number of impacted downstream customers and partners. The outage was cleared at around 2:30 AM EDT. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for March 4-10, 2024

ThousandEyes reported 142 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of March 4-10. That’s down 8% from 155 outages the week prior. Specific to the U.S., there were 63 outages, which is down 10% from 70 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages decreased from 95 to 91 outages, a 4% decrease compared to the week prior. In the U.S., the number of ISP outages stayed the same at 44 outages.

Public cloud network outages: Globally, cloud provider network outages fell from 13 to six outages. In the U.S., they decreased from seven to four outages.

Collaboration app network outages: Globally, collaboration app network outages decreased from eight outages to six. In the U.S., collaboration app network outages stayed at the same level as the week before: three outages.

Three notable outages

On March 5, several Meta services, including Facebook and Instagram, experienced a disruption that impacted users attempting to login, preventing them from accessing those applications. The disruption was first observed around 10:00 AM EST. During the disruption, Meta’s web servers remained reachable, with network paths to Meta services showing no significant error conditions, suggesting that a backend service, such as authentication, was the cause of the issue. The service was fully restored around 11:40 AM EST. More detailed analysis here.

On March 5, Comcast Communications experienced an outage that impacted a number of downstream partners and customers as well as the reachability of many applications and services, including Webex, Salesforce, and AWS. The outage, lasting 1 hour and 48 minutes, was first observed around 2:45 PM EST and appeared to impact traffic as it traversed Comcast’s network backbone in Texas, with Comcast nodes located in Dallas, TX and Houston TX, exhibiting outage conditions. The outage was completely cleared around 4:40 PM EST. More detailed analysis here.

On March 6, LinkedIn experienced a service disruption that impacted its mobile and desktop global user base. The disruption was first observed around 3:45 PM EST, with users experiencing service unavailable error messages. The major portion of the disruption lasted around one hour, during which time no network issues were observed connecting to LinkedIn web servers, further indicating the issue was application related. At around 4:38 PM EST, the service started to recover and was totally clear for all users around 4:50 PM EST. More detailed analysis here.

Additional details from ThousandEyes are available here.

Internet report for February 26-March 3, 2024

ThousandEyes reported 155 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of February 26-March 3. That’s down 6% from 165 outages the week prior. Specific to the U.S., there were 70 outages, which is up 19% from 59 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages decreased from 111 to 95 outages, a 14% decrease compared to the week prior. In the U.S., ISP outages increased 10%, climbing from 40 to 44 outages.

Public cloud network outages: After weeks of decreasing, cloud provider network outages began increasing again last week. Globally, cloud provider network outages climbed from eight to 13 outages. In the U.S., they increased from four to seven outages.

Collaboration app network outages: Globally, collaboration app network outages increased from five outages to eight. In the U.S., collaboration app network outages rose from two to three outages.

Two notable outages

On February 27, Level 3 Communications, a U.S. based Tier 1 carrier acquired by Lumen, experienced an outage that impacted multiple downstream partners and customers across the U.S. The outage, lasting a total of 18 minutes over a twenty-five-minute period, was first observed around 2:25 AM EST and appeared to be centered on Level 3 nodes located in Cleveland, OH. The outage was cleared around 2:50 AM EST. Click here for an interactive view.

On February 28, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across the U.S. The outage was first observed at around 2:00 PM EST and appeared to center on Time Warner Cable nodes located in New York, NY.  Five minutes into the outage, the number of nodes located in New York, NY, exhibiting outage conditions increased. The outage lasted 14 minutes and was cleared at around 2:15 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for February 19-25, 2024

ThousandEyes reported 165 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of February 19-25. That’s down significantly from 243 outages in the week prior – a decrease of 32%. Specific to the U.S., there were 59 outages, which is down 34% from 90 outages the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages decreased from 121 to 111 outages, an 8% decrease compared to the week prior. In the U.S., ISP outages decreased from 48 to 40 outages, a 17% decrease compared to the previous week.

Public cloud network outages: Globally, cloud provider network outages decreased significantly from 42 to eight outages, a 81% decrease compared to the week prior. In the U.S., they fell from eight to four outages.

Collaboration app network outages: Globally, collaboration app network outages decreased from seven outages to five. In the U.S., collaboration app network outages remained at the same level as the week prior: two outages.

Two notable outages

On February 22, Hurricane Electric, a network transit provider headquartered in Fremont, CA, experienced an outage that impacted customers and downstream partners across multiple regions, including the U.S., Australia, China, the U.K., Japan, Singapore, India, France, and Canada. The outage, first observed around 9:10 AM EST, lasted 32 minutes in total and was divided into two occurrences over a forty-five-minute period. The first occurrence appeared to initially center on Hurricane Electric nodes located in New York, NY, Phoenix, AZ and Indianapolis, IN. Ten minutes after appearing to clear, the nodes located in New York, NY, were joined by nodes located in San Jose, CA in exhibiting outage conditions. Five minutes into the second occurrence, the disruption appeared to radiate out, and the nodes located in New York, NY, Phoenix, AZ and Indianapolis, IN, were joined by nodes located in Seattle, WA, Denver, CO, Ashburn, VA, Kansas City, MO and Omaha, NE in exhibiting outage conditions. This increase in impacted nodes appeared to coincide with an increase in the number of impacted downstream customers and partners. The outage was cleared at around 9:55 AM EST. Click here for an interactive view.

On February 21, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across the U.S. The outage was first observed at around 2:45 PM EST and appeared to center on Time Warner Cable nodes located in New York, NY.  Fifteen minutes into the outage, the number of nodes located in New York, NY, exhibiting outage conditions increased. The outage lasted 23 minutes and was cleared at around 3:10 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for February 12-18, 2024

ThousandEyes reported 243 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of February 12-18. That’s down from 319 outages in the week prior – a decrease of 24%. Specific to the U.S., there were 90 outages, which is down slightly from 91 the week prior. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages decreased from 134 to 121 outages, a 10% decrease compared to the week prior. In the U.S., ISP outages decreased from 60 to 48 outages, a 20% decrease compared to the previous week.

Public cloud network outages: Globally, cloud provider network outages decreased significantly from 107 to 42 outages, a 61% decrease compared to the week prior. In the U.S., they doubled from four to eight outages.

Collaboration app network outages: Globally, collaboration app network outages decreased from 11 outages to seven. In the U.S., collaboration app network outages decreased from 5 to 2 outages.

Two notable outages

On February 16, Hurricane Electric, a network transit provider headquartered in Fremont, CA, experienced an outage that impacted customers and downstream partners across multiple regions, including the U.S., Egypt, Sweden, the U.K., Japan, Mexico, Australia, Argentina, the Netherlands, Belgium, and Canada. The outage, first observed around 8:25 AM EST, lasted 23 minutes in total and was divided into two occurrences over a thirty-minute period. The first occurrence appeared to initially center on Hurricane Electric nodes located in New York, NY. Fifteen minutes into the first occurrence, the nodes located in New York, NY, were joined by nodes located in Paris, France and Amsterdam, the Netherlands in exhibiting outage conditions.  Five minutes after appearing to clear, nodes located in New York, NY once again began exhibiting outage conditions. The outage was cleared at around 8:55 AM EST. Click here for an interactive view.

On February 17, AT&T experienced an outage on their network that impacted AT&T customers and partners across the U.S. The outage, lasting around 14 minutes, was first observed around 3:40 PM EST, appearing to center on AT&T nodes located in Little Rock, AR. Five minutes after first being observed, the number of nodes exhibiting outage conditions located in Little Rock, AR, appeared to rise. This increase in nodes exhibiting outage conditions appeared to coincide with a rise in the number of impacted partners and customers. The outage was cleared at around 3:55 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for February 5-11, 2024

ThousandEyes reported 319 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of February 5-11. That’s up from 265 outages in the week prior – an increase of 20%. Specific to the U.S., there were 91 outages. That’s up from 45 outages the week prior, an increase of 102%. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages increased from 106 to 134 outages, a 26% increase compared to the week prior. In the U.S., ISP outages more than doubled from 28 to 60 outages, a 114% increase compared to the previous week.

Public cloud network outages: Globally, cloud provider network outages decreased slightly from 117 to 107, a 9% decrease compared to the week prior. In the U.S., they decreased from five to four outages.

Collaboration app network outages: Globally, collaboration app network outages climbed from three outages to 11. In the U.S., there were five collaboration app network outages, up from zero the week prior.

Two notable outages

On February 7, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across multiple regions, including the U.S., Ireland, the U.K., Canada, India, Australia, Singapore, Japan, the Netherlands, France, Germany, Indonesia, Hong Kong, South Korea, China, and Brazil. The outage was observed across a series of occurrences over the course of forty-five minutes. First observed at around 4:50 PM EST, the outage, consisting of five equally spaced four-minute periods, appeared to initially center on Time Warner Cable nodes in New York, NY. Five minutes after appearing to clear, nodes located in New York, NY, were again observed exhibiting outage conditions, joined by nodes located in San Jose, CA. By the third period, the nodes located in San Jose, CA, had appeared to clear and were instead replaced by nodes located in Los Angeles, CA, in exhibiting outage conditions, in addition to nodes located in New York, NY. The outage lasted a total of 20 minutes and was cleared at around 5:35 PM EST. Click here for an interactive view.

On February 6, NTT America, a global Tier 1 ISP and subsidiary of NTT Global, experienced an outage that impacted some of its customers and downstream partners in multiple regions, including the U.S., Germany, the U.K., the Netherlands, and Hong Kong The outage, lasting 24 minutes, was first observed around 8:10 PM EST and appeared to initially center on NTT nodes located in Chicago, IL and Dallas, TX. Around five minutes into the outage, the nodes located in Chicago, IL and Dallas, TX, were joined by nodes located in Newark, NJ, in exhibiting outage conditions. The apparent increase of nodes exhibiting outage conditions appeared to coincide with an increase in the number of impacted downstream customers and partners. The outage was cleared around 8:35 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for January 29- February 4, 2024

ThousandEyes reported 265 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of January 29- February 4. That’s more than double the number of outages in the week prior (126). Specific to the U.S., there were 45 outages. That’s down from 55 outages the week prior, a decrease of 18%. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages was 106, an increase of 15% compared to 92 outages the previous week. In the U.S., ISP outages decreased by 28%, dropping from 39 to 28 outages.

Public cloud network outages: Globally, cloud provider network outages skyrocketed from five to 117 last week (the increase appeared to be a result of an increase in outages in the APJC region). In the U.S., they increased from two to five outages.

Collaboration app network outages: Globally, collaboration app network outages decreased from five outages to three. In the U.S., collaboration app network outages decreased from one outage to zero.

Two notable outages

On January 31, Comcast Communications experienced an outage that impacted a number of downstream partners and customers across multiple regions including the U.S., Malaysia, Singapore, Hong Kong, Canada, Germany, South Korea, Japan, and Australia. The outage, lasting 18 minutes, was first observed around 8:00 PM EST and appeared to be centered on Comcast nodes located in Ashburn, VA. Ten minutes into the outage, the nodes exhibiting outage conditions, located in Ashburn, VA, appeared to increase. The apparent increase of nodes exhibiting outage conditions appeared to coincide with an increase in the number of impacted downstream customers and partners. The outage was cleared around 8:20 PM EST. Click here for an interactive view.

On February 2, NTT America, a global Tier 1 ISP and subsidiary of NTT Global, experienced an outage that impacted some of its customers and downstream partners in multiple regions, including the U.S., Germany, the Netherlands, and the U.K. The outage, lasting 23 minutes, was first observed around 1:25 PM EST and appeared to center on NTT nodes located in Dallas, TX and Chicago, IL. The outage was cleared around 1:50 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for January 22-28, 2024

ThousandEyes reported 126 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of January 22-28. That’s down from 156 the week prior, a decrease of 19%. Specific to the U.S., there were 55 outages. That’s down from 91 outages the week prior, a decrease of 40%. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages was 92, a decrease of 14% compared to 107 outages the previous week. In the U.S., ISP outages decreased by 35%, dropping from 60 to 39 outages.

Public cloud network outages: Globally, cloud provider network outages dropped from 14 to five last week. In the U.S., they decreased from seven to two outages.

Collaboration app network outages: Globally, collaboration app network outages remained the same as the week prior: five outages. In the U.S., collaboration app network outages decreased from four outages to one.

Three notable outages

On January 26, Microsoft experienced an issue that affected its customers in various regions around the globe. The outage was first observed around 11:00 AM EST and seemed to cause service failures in Microsoft Teams, which affected the usability of the application for users across the globe. While there was no packet loss when connecting to the Microsoft Teams edge servers, the failures were consistent with reported issues within Microsoft’s network that may have prevented the edge servers from reaching the application components on the backend. The incident was resolved for many users by 6:10 PM EST. Click here for an interactive view.

On January 24, Akamai experienced an outage on its network that impacted content delivery connectivity for customers and partners using Akamai Edge delivery services in the Washington D.C. area. The outage was first observed around 12:10 PM EST and appeared to center on Akamai nodes located in Washington D.C. The outage lasted a total of 24 minutes. Akamai announced that normal operations had resumed at 1:00 PM EST. Click here for an interactive view.

On January 23, Internap, a U.S based cloud service provider, experienced an outage that impacted many of its downstream partners and customers in multiple regions, including the U.S., and Singapore. The outage, which was first observed around 2:30 AM EST, lasted 18 minutes in total and appeared to be centered on Internap nodes located in Boston, MA. The outage was at its peak around fifteen minutes after being observed, with the highest number of impacted regions, partners, and customers. The outage was cleared around 2:55 AM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for January 15-21, 2024

ThousandEyes reported 156 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of January 15-21. That’s up from 151 the week prior, an increase of 3%. Specific to the U.S., there were 91 outages. That’s up significantly from 63 outages the week prior, an increase of 44%. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages was 107, an increase of 8% compared to 83 outages the previous week, and in the U.S. ISP outages increased by 58%, climbing from 38 to 60 outages.

Public cloud network outages: Globally, cloud provider network outages dropped from 30 to 14 last week. In the U.S., they increased from six to seven outages.

Collaboration app network outages: Globally, collaboration app network outages decreased from seven to five outages. In the U.S., collaboration app network outages stayed at the same level: four outages.

Two notable outages

On January 16, Oracle experienced an outage on its network that impacted Oracle customers and downstream partners interacting with Oracle Cloud services in multiple regions, including the U.S., Canada, China, Panama, Norway, the Netherlands, India, Germany, Malaysia, Sweden, Czech Republic, and Norway. The outage was first observed around 8:45 AM EST and appeared to center on Oracle nodes located in various regions worldwide, including Ashburn, VA, Tokyo, Japan, San Jose, CA, Melbourne, Australia, Cardiff, Wales, London, England, Amsterdam, the Netherlands, Frankfurt, Germany, Slough, England, Phoenix, AZ, San Francisco, CA, Atlanta, GA, Washington D.C., Richmond, VA, Sydney, Australia, New York, NY, Osaka, Japan, and Chicago, IL. Thirty-five minutes after first being observed, all the nodes exhibiting outage conditions appeared to clear. A further ten minutes later, nodes located in Toronto, Canada, Phoenix, AZ, Frankfurt, Germany, Cleveland, OH, Slough, England, Ashburn, VA, Washington, D.C., Cardiff, Wales, Amsterdam, the Netherlands, Montreal, Canada, London, England, Sydney, Australia, and Melbourne, Australia began exhibiting outage conditions again.  The outage lasted 40 minutes in total and was cleared at around 9:50 AM EST. Click here for an interactive view.

On January 20, Hurricane Electric, a network transit provider headquartered in Fremont, CA, experienced an outage that impacted customers and downstream partners across multiple regions, including the U.S., Thailand, Hong Kong, India, Japan, and Australia. The outage, first observed around 7:15 PM EST, lasted 11 minutes in total and was divided into two occurrences over a one-hour five-minute period. The first occurrence appeared to center on Hurricane Electric nodes located in Los Angeles, CA. Fifty minutes after the first occurrence appeared to clear, the second occurrence was observed. Lasting 8 minutes, the outage initially appeared to center on nodes located in Los Angeles, CA. Around five minutes into the second occurrence, the nodes in Los Angeles, CA were joined by nodes located in San Jose, CA, in exhibiting outage conditions. The outage was cleared at around 8:20 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for January 8-14, 2024

ThousandEyes reported 151 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week of January 8-14. That’s up from 122 the week prior, an increase of 24%. Specific to the U.S., there were 63 outages. That’s up from 58 outages the week prior, an increase of 9%. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages was 83, an increase of 8% compared to the previous week, and in the U.S. they increased by 6%, climbing from 36 to 38 outages.

Public cloud network outages: Globally, cloud provider network outages jumped from 19 to 30 last week. In the U.S., they decreased from 10 to six outages.

Collaboration app network outages: Globally, collaboration app network outages increased from five to seven outages. In the U.S., collaboration app network outages increased from one to four outages.

Two notable outages

On January 14, Zayo Group, a U.S. based Tier 1 carrier headquartered in Boulder, Colorado, experienced an outage that impacted some of its partners and customers across multiple regions including the U.S., Canada, Sweden, and Germany. The outage lasted around 14 minutes, was first observed around 7:10 PM EST, and appeared to initially center on Zayo Group nodes located in Houston, TX. Ten minutes after first being observed, nodes located in Houston, TX, were joined by nodes located in Amsterdam, the Netherlands, in exhibiting outage conditions. This rise of the number of nodes exhibiting outage conditions appeared to coincide with an increase in the number of impacted downstream partners and customers. The outage was cleared around 7:25 PM EST. Click here for an interactive view.

On January 13, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across the U.S. The outage was first observed at around 12:45 PM EST and appeared to center on Time Warner Cable nodes located in New York, NY.  Fifteen minutes into the outage, the number of nodes located in New York, NY, exhibiting outage conditions increased. The outage lasted 19 minutes and was cleared at around 1:05 PM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

Internet report for January 1-7, 2024

ThousandEyes reported 122 global network outage events across ISPs, cloud service provider networks, collaboration app networks and edge networks (including DNS, content delivery networks, and security as a service) during the week ofJanuary 1-7. Over the prior three weeks, all outage categories continuously decreased for two weeks before increasing in the last week. Specific to the U.S., there were 58 outages. Here’s a breakdown by category:

ISP outages: Globally, the number of ISP outages was 77, an increase of 43% compared to the previous week, and in the U.S. they nearly doubled from 20 to 36.

Public cloud network outages: Globally, cloud provider network outages increased from 13 to 19 last week. In the U.S., they increased from 6 to 10.

Collaboration app network outages: Globally, collaboration app network outages increased from one to five outages. In the U.S., collaboration app network outages increased from zero to one. 

Two notable outages

On January 4, Time Warner Cable, a U.S. based ISP, experienced a disruption that impacted a number of customers and partners across the U.S. The outage was first observed at around 10:45 AM EST and appeared to center on Time Warner Cable nodes located in New York, NY.  Five minutes into the outage, the number of nodes located in New York, NY, exhibiting outage conditions increased. The outage lasted 13 minutes and was cleared at around 11:00 AM EST. Click here for an interactive view.

On January 4, Telecom Italia Sparkle, a Tier 1 provider headquartered in Rome, Italy, and part of the Italian-owned Telecom Italia, experienced an outage that impacted many of its downstream partners and customers in multiple regions, including the U.S., Argentina, Brazil, and Chile. The outage lasted 28 minutes in total and was divided into two episodes over a 35-minute period. It was first observed around 4:00 AM EST. The first period of the outage, lasting around 24 minutes, appeared to be centered on Telecom Italia Sparkle nodes located in Miami, FL. Five minutes after appearing to clear, nodes located in Miami, FL, again exhibited outage conditions. The outage was cleared around 4:35 AM EST. Click here for an interactive view.

Additional details from ThousandEyes are available here.

If you’ve never used bash case statements, you might be very surprised by how easily they allow you to simplify the logic of your scripts. Instead of composing a long group of code lines that would allow you to test a variable against a number of potential values, you could write a terse and delightfully readable chunk of code that’s easy to put together. When you use the if/then option, you have quite a bit of flexibility, but you can end up having to put a lot of code together. For example, you could test a variable and take an action based on its value in a script like this:

#!/bin/bash

echo -n "enter the number of equal sides that the shape has> "
read numsides

# make sure answer is numeric
re='^[0-9]+$'
if ! [[ $numsides =~ $re ]] ; then
   echo "error: Not a number" >&2; exit 1
fi

if [ $numsides -lt 3 ]; then
    echo "Sorry, but that would NOT be a geometric shape"
fi
if [ $numsides == 3 ]; then
    echo triangle
fi
if [ $numsides == 4 ]; then
    echo square or rectangle
fi
if [ $numsides == 5 ]; then
    echo pentagon 
fi
if [ $numsides == 6 ]; then
    echo hexagon 
fi
if [ $numsides == 7 ]; then
    echo heptagon 
fi
if [ $numsides == 8 ]; then
    echo octagon 
fi
if [ $numsides == 9 ]; then
    echo nonaagon 
fi
if [ $numsides == 10 ]; then
    echo decagon 
fi
if [ $numsides == 11 ]; then
    echo hendecagon 
fi
if [ $numsides == 12 ]; then
    echo dodecagon 
fi
if [ $numsides -gt 12 ]; then
    echo "Hmm, you’d better ask Google"
fi

Notice that the script will exit with an error if the value entered isn’t numeric or is not provided.

You can also take advantage of “elif” (else if) clauses to lessen the number of lines you need. Here’s a redo of the script above that displays this option.

#!/bin/bash

echo -n "enter the number of equal sides that the shape has> "
read numsides

# make sure answer is numeric
re='^[0-9]+$'
if ! [[ $numsides =~ $re ]] ; then
   echo "error: Not a number" >&2; exit 1
fi

if [ $numsides -lt 3 ]; then
    echo "Sorry, but that would NOT be a geometric shape"
elif [ $numsides == 3 ]; then
    echo triangle
elif [ $numsides == 4 ]; then
    echo square or rectangle
elif [ $numsides == 5 ]; then
    echo pentagon 
elif [ $numsides == 6 ]; then
    echo hexagon 
elif [ $numsides == 7 ]; then
    echo heptagon 
elif [ $numsides == 8 ]; then
    echo octagon 
elif [ $numsides == 9 ]; then
    echo nonagon 
elif [ $numsides == 10 ]; then
    echo decagon 
elif [ $numsides == 11 ]; then
    echo hendecagon 
elif [ $numsides == 12 ]; then
    echo dodecagon 
elif [ $numsides -gt 12 ]; then
    echo "Hmm, you’d better ask Google"
fi

Using case statements, on the other hand, makes your code much simpler to put together and much easier to read.

#!/bin/bash

echo -n "enter the number of equal sides that the shape has> "
read numsides

# make sure answer is numeric
re='^[0-9]+$'
if ! [[ $numsides =~ $re ]] ; then
   echo "error: Not a number" >&2; exit 1
fi

case "$numsides" in
    0-2) echo "Sorry, but that would NOT be a geometric shape";;
    3) echo triangle;;
    4) echo square or rectangle;;
    5) echo pentagon;;
    6) echo hexagon;;
    7) echo heptagon;;
    8) echo octogon;;
    9) echo nonagon;;
    10) echo decagon;;
    11) echo hendecagon;;
    12) echo dodecaggon;;
    *) echo "Hmm, you’d better ask Google"
esac

Each segment in the case statement must end with two semicolons. However, you could use multiple commands separated by semicolons and only use the double semicolons at the end od that segment.

Wrap-up

Notice how much easier case commands are to set up and maintain. You can provide any number of possible values at the cost of one line each.

Intel managed to slow the momentum of AMD’s ascent for the first quarter of 2024. But year-over-year, AMD is still growing.

The first quarter is traditionally a slow one for chip sales, coming off the holiday rush. That applies both to clients and servers, but for different reasons. Client CPU sales come from holiday giving, while server CPU sales in the fourth quarter tend to be end-of-the-year spending. People have budgets to spend by the end of the year, so there is often a rush in the fourth quarter for server sales.

For the first quarter of 2024, x86 processor shipments were finally back in line with typical seasonal trends for both client and server processors, according to Mercury Research. This past quarter was the first time since early 2020, when the Covid pandemic began, that market results have been completely normal in the desktop, mobile and server segments, wrote Dean McCarron, principal analyst with Mercury Research.

IoT/SoC processor shipments declined far more than typical, as decreased console demand impacted AMD’s SoC shipments, said AMD CEO Lisa Su in an earnings call with Wall Street analysts. Su noted that the consoles on the market are about five years old, and sales are beginning to slack off.

McCarron said the SOC weakness wasn’t the only thing dragging down AMD; he noted some softness in its mobile processor sales versus Intel as well.

Much of the past year has been spent on inventory correction as chip vendors draw down their excessive CPU inventory and make room for new products. It will be at least two more quarters before a proper comparison of sales can be made.

So, for the quarter, Intel’s share of the x86 market was 73.9%, a slight bump from the 71.4% in Q4 of 2023 and a considerable rise from 65.4% in Q1 of 2023. AMD’s share was 26.1%, down 20.6% sequentially and 34.6% year-over-year.

Again, this is attributable to weakness in the gaming console industry. Intel makes a SOC chip that is used in both the Sony PlayStation and Apple Xbox – which is rather ironic, because those sales were what kept AMD afloat several years ago before its revival.

When you take out the SOC and IoT businesses, there is a slight change in favor of AMD. For Q1 of 2004, Intel held a 79.2% share to AMD’s 20.8%. In the prior quarter, it was 79.6% to 20.4%, reflecting no substantive change from one quarter to the next. A year prior, it was 82.8% versus 17.2%, so AMD’s gains came earlier in the year.

In the server field, things were pretty much unchanged sequentially, with Intel holding a 76.4% share to AMD’s 23.6%. In the prior quarter, it was 76.9% to 23.1%. But for the year prior, it was 82% versus 18%, reflecting a positive reception of the fourth-generation Epyc server processors, launched in November 2022.

McCarron wrote that server CPU sales took a pretty hard hit in 2023, and Q4 is the only quarter to show sequential growth. He said that judging by the talk on the earnings call, it is apparent that Genoa is the main driver of on-year growth for AMD.

McCarron also noted that AMD shipped its MI300A Instinct processor, which is a hybrid CPU GPU accelerator, in the first quarter. Had these units been counted as CPUs, AMD’s on-quarter share would have been higher.

Today’s enterprises need more help managing and securing their distributed networking environments than they ever have, Cisco concludes in its Global Networking Trends Report. The research, released this week, analyzes the networking challenges, IT and business priorities, architectural maturity, and investment strategies of 2,052 IT professionals across 10 global industries.

“Network architectures are more sophisticated, more complex, and spread across more multi-clouds and multi-vendors than ever. IT leaders are also besieged by rising cybersecurity risks, increased demand from new app and workload types, and vastly distributed workforces and infrastructures,” wrote Jonathan Davidson, executive vice president and general manager of Cisco Networking, in a blog about the study. 

Managing and controlling such widely distributed systems are core challenges enterprise network customers face, Davidson stated.

One in five (21%) organizations surveyed are currently using multiple, separate management systems when managing their campus, branch, WAN, data center, and multicloud domains, according to the report, while 39% are leveraging a platform architecture across some networking domains. Using a centralized network platform to help handle new and future environments could resolve those problems, according to Cisco’s data.

Cisco defines a network platform as an integrated system that combines software, policy, and open APIs with an intuitive user interface, advanced telemetry, and automation. It provides a centralized operator experience, end-to-end network management, and an API-driven ecosystem that simplifies operations and enables digital experiences across one or more networking domains. According to the report:

“Adoption of the network platform is anticipated to increase, with nearly three-quarters (72%) of respondents expecting to leverage a platform architecture across one or more domains. Of those, 39% expect to scale the platform architecture across all networking domains, greatly simplifying network management and enabling a variety of business, operational, and technical benefits – from open API ecosystems to enhanced IT collaboration across functions, easier integrations, and a rich pool of data. Data aggregated from telemetry across network domains provides an enterprise-wide view that can be used to generate insights from AI analytics, enhancing performance, experience, security, cost efficiency, and sustainability.”

2024 Cisco Global Networking Trends Report

As networks scale to handle new applications in the data center and cloud, the threat landscape grows, Davidson stated. 

According to the report, 40% of IT leaders cite cybersecurity risks as their number one concern impacting network strategy over the next 12 months, and respondents said they were handling this challenge in a couple ways, according to Davidson: “First is integrating network and security processes, technology, and tools, with half of respondents making this their top network security investment over the next two years. Second, is moving more security tools to the cloud to protect the increasingly distributed infrastructure and workforce better.”

When asked about their top network security priorities for the next two years, 52% said the integration of network security into broader IT security functions was their top priority. To that end, 76% of those organizations plan to deploy a secure access service edge (SASE) architecture with integration of SD-WAN and security service edge (SSE) cloud security within the same timeframe, researchers stated.

In addition, 44% of respondents said that faster cybersecurity threat identification and response is the most important benefit expected from the convergence of networking and security technologies, processes, and tools.

“Sharing data and telemetry between networking and security domains was the second ranked benefit for 29% of respondents, followed by providing consistent, secure access to multicloud applications from anywhere for 27% of respondents,” the report stated.

Multicloud environments are prevalent, with 92% of organizations saying they have deployed two or more public cloud providers to host their workloads and 34% using more than four, according to last year’s networking trends report.

“However, each public cloud service provider, private data center, and hybrid cloud environment uses different network and security operational models. Organizations need to address the resulting management complexity with a strategy that enables better visibility and more consistent control of connectivity and security across disparate private and public cloud environments,” researchers stated.

Looking ahead to two years from now, 60% of companies expect to have an integrated multicloud networking and security management platform with common APIs for secure workload mobility, network and application visibility, and policy management, researchers stated. “Additionally, organizations on the leading edge of IT innovations plan to have AIOps-driven automation in place to optimize workload mobility with end-to-end visibility across multicloud, private cloud, and edge networks,” the report found.

On the AI front, there’s the potential for AI to relieve IT teams that lack enough operational support. “The promise of AI is the needed reprieve for IT organizations struggling with a lack of resources and automation to handle basic operational tasks,” Davidson stated. “Only 5% of respondents believe their teams are equipped to deliver the innovations needed to help steer business strategy, satisfy customers, and optimize operations.”

Within two years, 60% of respondents expect AI-enabled predictive automation across all domains to manage and simplify network operations.

Data center upgrades are on tap to enable greater throughput and scalability required for AI workloads; 56% of respondents are planning to deploy Enhanced Ethernet, for example, and 59% of respondents plan to simplify their data center network operations with AIOps within two years, the report stated.

Some other interesting study findings include:

• Today, 32% of organizations said InfiniBand is currently the technology of choice to connect high performance workloads. In two years, 56% said they will deploy a next generation enhanced Ethernet network to support AI workloads.
• Today, 42% of organizations said sustainability is the top area where the network can drive digital business success. In two years, 55% feel IT will be leading their organizations’ sustainability strategy, and they expect network protocol-driven sustainability initiatives like selective powering and the use of low power modes to positively impact sustainability goals.
• Today, 32% of respondents said they don’t have visibility into their IT-related energy consumption or emissions data. But in two-years, 60% expect network-driven energy management solutions to have a significant, positive impact on sustainability strategy.

In a reversal of earlier reports, Amazon Web Services (AWS) has clarified that it has not halted orders for Nvidia’s most advanced chip. Instead, AWS and Nvidia are collaborating on a project-specific upgrade, a Financial Times report said.

The Financial Times initially reported that AWS has “halted” its orders for Nvidia’s Grace Hopper chip. But now the report said that AWS has now “fully transitioned” it to the newest Blackwell GPUs, announced in March. Some interpreted this transition as a potential halt in orders for the current generation chips.

The transition applies solely to Project Ceiba, a joint supercomputer initiative between AWS and Nvidia, said the report, quoting an AWS spokesperson.

“To be clear, AWS did not halt any orders from Nvidia,” the Financial Times report said quoting the AWS spokesperson. “In our close collaboration with Nvidia, we jointly decided to move Project Ceiba from Hopper to Blackwell GPUs, which offer a leap forward in performance.”

Project Ceiba refers to an AI supercomputer being co-developed by AWS and Nvidia.

The Financial Times has since updated its story to reflect that Amazon’s chip orders had not yet been placed, aligning with AWS’ clarification.

AWS continues to offer services based on Nvidia’s Hopper chips, which remain crucial for training AI systems. The decision to transition Project to Ceiba to Blackwell chips aligns with Nvidia’s announcement in March, highlighting the superior performance of the new GPUs.

Blackwell promises a performance boost

Nvidia’s new Blackwell chips, unveiled by CEO Jensen Huang in March, are expected to be twice as powerful for training large language models (LLMs) such as OpenAI’s ChatGPT, compared to their predecessors.

The Nvidia GB200 Grace Blackwell Superchip integrates two NVIDIA B200 Tensor Core GPUs with the Nvidia Grace CPU via a 900GBps ultra-low-power NVLink chip-to-chip interconnect.

To achieve the highest AI performance, GB200-powered systems can be paired with the newly announced Nvidia Quantum-X800 InfiniBand and Spectrum-X800 Ethernet platforms, which offer advanced networking capabilities at speeds up to 800Gbps.

“Selecting the flagship Blackwell chips in lieu of less powerful Grace Hoppers from Nvidia makes more sense for advancing its AI training of LLMs, LVMs, and simulation with applications across industries,” said Neil Shah, VP for research and partner at Counterpoint Research. “For AWS, especially with this rapid evolution of the size of the training models, the cloud giant has to be prudent of its investments in getting the best ROI for the advanced compute investments as well as the efficiency of those compute from an energy consumption perspective. With Project Cebia, the goalposts are actually moving and Amazon needs to be at the leading edge to catch up with Google, and Microsoft in this AI race.”

AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure will be among the first cloud service providers to offer Blackwell-powered instances, Nvidia said in its March announcement. Additionally, companies in the Nvidia Cloud Partner program, including Applied Digital, CoreWeave, Crusoe, IBM Cloud, and Lambda, will also provide these advanced instances.

IT observability and management software maker SolarWinds today introduced a new AI platform that when embedded into the company’s other solutions will provide generative AI capabilities to speed problem resolution, minimize downtime, and empower IT staff to improve service delivery.

SolarWinds AI, developed using SolarWinds’ AI by Design framework, is embedded in SolarWinds AI in Service Desk, an IT service management (ITSM) solution. SolarWinds AI in Service Desk uses large language models and proprietary algorithms to help service agents by summarizing complex ticket histories, providing suggested agent responses to inquiries, and generating real-time recommended steps for resolving issues, according to the company.

“In creating SolarWinds AI, we leveraged our long history of engaging with the IT community to build an AI system that could act as a valued partner to IT pros in their everyday lives and reimagine the productivity, efficiency, and innovation they’re capable of,” said Krishna Sai, senior vice president of engineering at SolarWinds, in a company statement.

SolarWinds AI in Service Desk uses both traditional AI and generative AI to help agents more quickly resolve issues and close tickets. AI helps with the automatic routing and classification of incoming incidents, according to SolarWinds, which reduces the impact of ticket misclassification and human error in the process. For service portal users—for instance, end users—AI will suggest knowledge base articles and service catalog items during the creation of a new incident, steering end users in the right direction toward resolution. Virtual agents also use AI to mimic human interaction and understand the end user’s language, asking additional questions to gain context, responding to queries in real time, and handing off to a human agent when necessary.

SolarWinds’ Sai first discussed AI by Design in a blog post that detailed the four principles upon which it is based. “AI by Design is crafted to ensure that our customers can leverage the very best AI has to offer while minimizing potential downsides. Crucially, it isn’t a static framework but a dynamic and evolving set of guidelines that will adapt to emerging insights,” Sai wrote in the SolarWinds blog post.

According to the company, the principles applied in SolarWinds AI are:

• Privacy and security: SolarWinds AI is designed to secure personal data collection, storage, use, and protection using role-based access control (RBAC), multi-factor authentication (MFA), privileged access management (PAM), and anonymization techniques. Sensitive customer data does not pass through external large language models (LLMs), and SolarWinds AI improves based on usage patterns and organizational needs without compromising data privacy.

• Accountability and fairness: The framework is designed to include humans in training and testing to oversee AI-guided decision-making processes to avoid biases. It captures and addresses negative user experiences through feedback and validation mechanisms, ensuring that human judgment plays a role throughout the AI lifecycle.

• Transparency and trust: SolarWinds creates an “explainability pipeline” to demonstrate the rationale behind AI-generated actions. This enables continuous refinement based on the customers’ needs and demystifies the process for AI decisions and outputs.

• Simplicity and accessibility: Ease-of-use principles are built into the company’s product interfaces to help ensure the continuation of regular workflows and existing user procedures, behaviors, and practices.

SolarWinds plans in the future to incorporate SolarWinds AI throughout its suite of observability and IT management solutions.  SolarWinds offers both perpetual and subscription-based licenses for on-premises and SaaS-based solutions.

Broadcom has introduced a new series of 400G Ethernet adapters specifically tuned for resolving network bottlenecks when moving massive amounts of data around for AI processing. The new processors are equipped with a third-generation RoCE pipeline, low-latency congestion control technology, and telemetry functions. They are meant for a high-bandwidth, high-stress network environment associated with AI infrastructure.

But don’t call it a smartNIC. Jas Tremblay, vice president and general manager of the Data Center Solutions Group at Broadcom, said there is a difference between their product and what is generally accepted as a smartNIC.

“These are traditional performance NICs optimized for AI,” he said. “They don’t have a large, multi-core CPU running on them. They’re fully hardware offload, high performance, optimized for low power AI. This processor will not do application offload on it. It’s really [about] connectivity, high performance, low latency, congestion control.”

The processors are not only available as standard firmware installed on a network card, but also as a chiplet option for chiplet designs – which Tremblay said is an industry first.

What this means, in theory, is that a server processor that uses a chiplet SoC design could integrate the Broadcom networking chip into their SOC. For example, AMD uses a chiplet design with its Epyc server processors, combining chips that contain cores, memory, and I/O on a single silicon die. If AMD wanted to, it could incorporate Broadcom’s networking technology into the Epyc chip die. That’s not to say AMD is doing that – but it has the option to do so.

There is a difference between the standalone chip and the chiplet, however; the standalone chip is more feature-rich with support for many different types of virtual machines, for example, while the chiplet is more performance optimized. Customers can choose what suits their needs.

Separately, Broadcom is coming out with an array of Ethernet adapters specifically designed to handle copper wiring instead of fiber and transmit the data over five meters in length. For all of its positives, fiber-optic is extremely sensitive to heat, and AI servers are virtual ovens with their immensely hot GPUs running at full utilization. So copper is a better choice for an AI server, and Broadcom has increased the length 400G data can travel.

The 400G PCIe Gen 5.0 Ethernet family (BCM57608) is broadly available from multiple server vendors, as well as from Broadcom.

At its Build 2024 event in Seattle this week, Microsoft released updates to its cloud infrastructure to bolster support for all workloads, including AI-related ones. The updates includes new Azure virtual machines (VM), a new provisioning service, and options for enabling access to Microsoft Copilot in Azure.

In the VM space, Microsoft shared plans to release Azure VMs that can run on Cobalt 100 processors currently in preview. The Azure Cobalt 100 CPU, which is built on Arm architecture, was launched in November last year by Microsoft in an attempt to make its infrastructure across data centers more energy efficient when compared to commercial AMD and Intel CPUs.

“Cobalt 100-based VMs are Azure’s most energy efficient compute offering and deliver up to 40% better performance than our previous generation of Arm-based VMs…. the new Cobalt 100 based VMs are expected to enhance efficiency and performance for both Azure customers and Microsoft products,” Omar Khan, general manager of Azure infrastructure marketing at Microsoft, said in a statement.

Additionally, IC3, the platform that powers customer conversations in Microsoft Teams, is adopting Cobalt to serve its growing customer base more efficiently, achieving up to 45% better performance on Cobalt 100 VMs, Khan added.

Alongside the Azure VMs for Cobalt CPUs, Microsoft made AMD’s ND MI300X series of processors generally available for Azure. The ND MI300X VM, which combines eight AMD MI300X Instinct accelerators, will provide enterprises with better cost performance than rivals, especially for inferencing large language models, such as GPT-4, according to Khan.

Microsoft also released a new provisioning service, dubbed Azure Compute Fleet, which simplifies provisioning of Azure compute capacity across different VM types, availability zones, and pricing models. This will enable enterprises to meet desired scale, performance, and cost by enabling users to control VM group behaviors automatically and programmatically, Khan explained.  

Other updates include opening up Copilot in Microsoft Azure to all enterprise customers over the next couple of weeks. “With this update, enterprise customers can choose to have all their users access Copilot or grant access to specific users or groups within a tenant,” Khan said.

Copilot in Azure was introduced to help enterprise teams manage cloud and edge operations in natural language.

“We want to inspire you to reimagine your organization for artificial intelligence, and we want to encourage you to act fast,” Dell Technologies CEO Michael Dell told attendees at the company’s customer event in Las Vegas on Monday.

Dell unveiled a number of new AI-related products at the event, including Dell AI Factory, with which the company aims to give its customers access to what it considers “the broadest AI portfolio in the industry.” This includes everything from devices to data centers and cloud, as well as an open ecosystem of technology partners to create AI applications through a traditional purchase or as a subscription to Dell Apex, its collection of multi-cloud services.

“AI is transforming business at an unprecedented rate. Data centers must be designed from the ground up to handle the speed and scale of AI, while new AI-powered PCs are transforming productivity and collaboration,” said Dell COO Jeff Clarke. “What’s needed is new IT infrastructure and devices designed specifically to meet the specific demands of AI.”

“The Dell AI Factory helps customers accelerate AI adoption with the world’s broadest AI portfolio and leading AI ecosystem partners, offering right-sized approaches and greater control over AI deployments on-premises, at the edge and across cloud environments,” Clarke said.

Dell’s end-to-end AI portfolio, spanning client devices, servers, storage, data protection and networking, forms the foundation of the Dell AI Factory.

Dell is expanding that portfolio with new offerings including Copilot+ PCs, PowerScale F910 all-flash file storage, an AI data protection solution, and the Dell PowerSwitch Z9864F-ON.

Speakers also addressed Dell’s partnerships with other vendors to provide integrated solutions to help customers get started with AI, including one with Nvidia to accelerate adoption of AI with full-stack solutions.

At the event, Dell also presented advancements in its PowerStore flash storage systems to improve the performance, efficiency, resiliency and mobility of multi-cloud data. In addition, Dell also announced the expansion of the Dell Apex portfolio with new advancements in AIOps and improved management of multi-cloud storage and Kubernetes.

“Our advancements in PowerStore and our new range of financial and operational benefits for customers and partners make a big impact and raise the bar for all-flash storage,” said Arthur Lewis, president of the firm’s Infrastructure Solutions group.

Dell PowerStore helps manage growing workload demands with more flexible four-tier cell storage (QLC) and significant performance improvements. Specifically, QLC-based storage delivers enterprise-class performance at a lower cost per terabyte compared to triple-tier cell (TLC) models. In addition, it boosts hardware performance by up to 66% through new upgrades of higher model devices.

“Our focus on innovation also extends to Dell Apex, where we are improving infrastructure and application reliability through the power of AI and automation, making it even easier to manage multicloud storage and Kubernetes,” Lewis said.

The company’s Apex AIOps software-as-a-service (SaaS) optimizes the health of Dell’s infrastructure and service availability with a significant expansion of AIOps tools that simplifies operations, improves IT agility, and offers greater control over applications and infrastructure through three integrated capabilities: infrastructure observability, application observability, and incident management. Dell Apex Navigator SaaS expands to include Kubernetes storage management and additional support for Dell Apex storage offerings for public cloud.

Three teams – in Boston, in China, and the Netherlands – have simultaneously announced that they’ve figured out ways to store entangled photons without breaking the entanglement, a critical step in building quantum repeaters, and, thus, scalable quantum networks.

The Boston team used silicon atoms embedded in diamond chips. A team in the Netherlands also used diamond crystals, but with nitrogen atoms instead of silicon. And the Chinese team used clouds of rubidium atoms. The American and Chinese teams both published their papers in this month’s Nature magazine, while the Netherlands research is available as a pre-print.

This is a big deal for two reasons. First, because it brings us closer to actually having secure quantum networks. And, second, because China is finally getting some competition in the quantum networking space, where it’s long held a substantial lead.

According to a McKinsey report released last month, China is far ahead in investment in quantum technology, with public investments totaling over $15 billion. The United States, at just $3.8 billion, is far behind.

China is also in the lead when it comes to the size of its quantum networks. In 2021, China built a 4,600-kilometer network between Beijing, Hefei, and Shanghai, an order of magnitude larger than networks elsewhere.

And, at the end of 2023, it demonstrated a secure quantum connection with Moscow, for a distance of 3,800 kilometers, according to a report in the South China Morning Post.

However, if traditional fiber is used to transmit the protons, there’s a limit to how far they can go. At long distances, the loss of photons due to noise in the fiber is exponential, says Mihir Bhaskar, senior research scientist and head of the AWS Center for Quantum Networking. Even with ideal fiber under laboratory conditions, single-hop trips top out at 40 to 50 kilometers, he says.

China’s 4,600 kilometer network uses trusted nodes to bridge longer distances, he says, meaning the entanglement gets broken after every leg of the trip. China also uses satellite communications, since photons can go very far in empty space.

But, until now, scalable, secure quantum networks that use existing infrastructure were impossible, he says. “We want to show that this technology is robust and can work over any kind of fiber.”

The ideal situation is to be able to transmit the photons along standard commercial fiber, with repeaters along the way that pass along the photons without breaking the entanglement. Now, it looks like this problem has been cracked.

In the Boston project, researchers from Harvard University and the AWS Center for Quantum Networking were able to use 35 kilometers of commercial fiber running under and above the streets of Boston to transmit entangled protons.

The researchers used a single silicon atom embedded in a diamond chip to store an entangled photon for up to a second before passing it along on the next leg of its journey. The chip doesn’t amplify the photon – doing so would break the entanglement – but it breaks up a long trip into short, more doable hops. “We store it in memory, and it’s just a buffer that waits until it’s able to establish the connection,” says Bhaskar. But we’re still five to ten years away, before the technology is commercially viable, he adds.

One problem is that each quantum repeater needs to be supercooled and is about the size of a large refrigerator. That’s not practical for commercial deployments.

Second, the repeater can currently handle just one photon at a time. For practical deployments, each repeater will need to handle multiple entangled photons at once. “The goal is to scale it up so we have lots of memories living on that diamond chip,” says Bhaskar. “Then you can support lots of engineering bandwidth.”

Finally, there’s the problem of business demand.

There are two main near-term use cases for quantum networks. The first use case is to transmit encryption keys. The idea is that public key encryption – the type currently used to secure Internet traffic – could soon be broken by quantum computers. Symmetrical encryption – where the same key is used to both encrypt and decrypt messages – is more future proof, but you need a way to get that key to the other party. Quantum key distribution uses the secure quantum network to send the key, then transmits the body of the communication via traditional methods. Quantum networks aren’t expected to have enough capacity to carry significant amounts of traffic anytime in the foreseeable future, but using them to send just the key is a practical application, especially for particularly sensitive government or financial communications.

Today, however, the encryption we currently have is good enough, and there’s no immediate need for companies to look for secure quantum networks. Plus, there’s progress already being made on creating quantum-proof encryption algorithms.

The other use for quantum networks is to connect quantum computers. Since quantum networks transmit entangled photons, the computers so connected would also be entangled, theoretically allowing for the creation of clustered quantum computers that act as a single machine.

“There are ideas for how to take quantum repeaters and parallelize them to provide very high connectivity between quantum computers,” says Oskar Painter, director of quantum hardware at AWS. “Or to provide private access to calculations performed by a quantum computer in the cloud.”

But, again, we don’t yet have commercially useful quantum computers on the market, so there’s not much demand in connecting them together, or connecting them to customers.

“We have our own internal efforts to build quantum computing hardware,” says Painter. “And we’re working on quantum networking solutions like these quantum repeaters. But both of those are research and development projects at this point – we foresee a number of years before these technologies become practical and we can bring them to customers.”

But when that day comes, being able to leverage existing fiber networks will be crucial, he says.

“People have built all kinds of quantum networks before, but we’ve never shown that those are compatible with real-world infrastructure,” adds Bhaskar. “What’s new here is that yes, we can do this. That’s a step forward in thinking about how we’re going to do this stuff over long distances, on a global scale, and in real data centers.”

Phishing threats have reached unprecedented levels of sophistication in the past year, driven by the proliferation of generative AI tools. Transforming how cybercriminals operate, AI advancements are revolutionizing and reshaping the phishing threat landscape. Moreover, this technology has democratized the ability to orchestrate intricate phishing campaigns, making it easier than ever for even beginners to conduct complex and believable phishing attacks. Specifically, this observed shift is enabling novice cybercriminals to launch highly convincing, personalized scams with ease. As a result, organizations now face a myriad of new challenges in protecting their data and systems from the increasing onslaught of phishing attacks.

In response, the Zscaler ThreatLabz team has released the 2024 Phishing Report. This report analyzes over 2 billion phishing transactions from 2023, found within the Zscaler cloud, to equip organizations with a clear understanding of the rapidly evolving phishing landscape. Providing insights into the latest trends and tactics used by cybercriminals, the report highlights active phishing campaigns, exposes emerging schemes, and identifies top targets by region, industry, imitated brand, and more. Showcasing real-world examples, ThreatLabz phishing findings underscore the importance of applying constant vigilance and zero trust security strategies. The guidance offered aims to help organizations strengthen their defenses against these evolving phishing techniques.

Download the Zscaler ThreatLabz 2024 Phishing Report to gain the knowledge needed to proactively combat the rising wave of new phishing threats.

6 key phishing findings

The following findings represent a subset of key phishing trend discoveries that shed light on the evolution of phishing tactics.

Top phishing trends

• Phishing attacks surged by 58.2% in 2023 compared to the previous year, reflecting the growing sophistication and reach of threat actors.
• Voice phishing (vishing) and deepfake phishing attacks are on the rise as attackers harness generative AI tools to amplify their social engineering tactics.
• Adversary-in-the-middle (AiTM) phishing attacks persist and browser-in-the-browser (BiTB) attacks are emerging as a growing threat. 

Top phishing targets

• The US, UK, India, Canada, and Germany were the top five countries targeted by phishing attacks.
  
    src="https://b2b-contenthub.com/wp-content/uploads/2024/05/Picture1_b372e9.png" alt="" loading="lazy" width="400px">
• The finance and insurance industry faced 27.8% of overall phishing attacks, marking the highest concentration among industries and a 393% year-over-year increase.
  src="https://b2b-contenthub.com/wp-content/uploads/2024/05/Picture2_5848c9.png" alt="" loading="lazy" width="400px">
   
• Microsoft remains the most frequently imitated brand, with 43.1% of phishing attempts targeting it.
  src="https://b2b-contenthub.com/wp-content/uploads/2024/05/Picture3_1c358f.png" alt="" loading="lazy" width="400px">
   

Discover further insights into each of these findings and more in the report.

Spotlight on AI-enabled phishing threats

GenAI has undoubtedly proven transformative in turning up productivity across businesses. Yet on the flip side of this transformation is a perilous truth: AI is also turning novice to average threat actors into skilled social engineers and sophisticated phishing attackers.

By automating and personalizing various components of the attack process, AI speeds up and refines phishing attacks, making them more sophisticated and difficult to detect.

• GenAI quickly analyzes public data, such as information about organizations and executives, saving time in reconnaissance for threat actors and enabling more precise targeted attacks.
• LLM chatbots craft accurate, believable phishing communications and emails by eliminating misspellings and grammar mistakes.
• GenAI can swiftly generate convincing phishing pages. The ThreatLabz report showcases how ChatGPT created a phishing login page in less than 10 prompts and provides key indicators to look out for when identifying a phishing page.

AI has blurred the line between authentic and fraudulent content, making it all the more challenging to discern phishing schemes from legitimate web pages and digital communication.

As ThreatLabz researchers tracked phishing trends throughout 2023, several notable advanced AI tactics also emerged. Among these were the rise of vishing and deepfake phishing, increasingly favored social engineering tactics that use AI-powered impersonation tools.

Vishing insights

Advanced vishing campaigns are gaining popularity globally, leading to substantial financial losses in some cases. In a notable attempt that ThreatLabz thwarted during the summer of 2023, phishing attackers used AI technology to perpetrate a vishing attack by impersonating Zscaler CEO Jay Chaudhry. The report details the sequence of events, serving as a critical reminder for enterprises and employees to stay vigilant against vishing scammers. ThreatLabz anticipates a continued surge in targeted voice phishing campaigns led by groups like Scattered Spider in the next year. As these efforts aim to acquire employee login credentials, it is imperative for organizations to fortify their phishing defenses to prevent unauthorized access and exploitation.

Deepfake insights

Phishing attacks involving deepfakes will be one of the most challenging AI-driven cyberthreats. Threat actors now possess the ability to create video content that precisely and accurately replicates faces, voices, and mannerisms. This manipulation has already manifested in concerning ways, such as in the electoral process, where deepfake videos fabricate false narratives or statements from political figures. These videos can sway public opinion, disseminate disinformation, and erode trust in the integrity of the electoral process. As society becomes more and more reliant on digital communication and media consumption, the potential political and life-altering ramifications of deep fake scams will likely extend far beyond the scope of current applications. From financial scams to corporate espionage, the use of deepfake technology poses a significant threat to organizations, individuals, and society at large.

Additionally, ThreatLabz observed a rise in QR code scams, recruitment scams, browser-in-the-browser (BitB) attacks, and adversary-in-the-middle (AiTM) attacks. Learn more about each of these schemes in the report.

Mitigate phishing risk with zero trust

Given the concerning threat landscape uncovered by this year’s report, how can organizations protect against the latest phishing threats? One definitive solution lies in establishing a foundation of a zero trust architecture. Adapting security strategies to combat new phishing trends and mitigate associated risks is crucial—and zero trust is a proven strategy.

The Zscaler ThreatLabz 2024 Phishing Report provides essential guidance to this end, including:

• Fighting AI with AI: Learn about Zscaler’s AI-powered phishing prevention capabilities needed to combat AI-driven threats, including preventing browser exploitation from phishing pages with Zscaler Browser Isolation
• Zero trust architecture advantages: Learn how the Zscaler Zero Trust Exchange prevents traditional and AI-driven phishing at multiple stages of the attack chain:

• Prevent compromise: TLS/SSL inspection at scale, AI-powered browser isolation, and policy-driven access controls prevent access to suspicious websites.
  • Eliminate lateral movement: Users connect directly to applications, not the network, while AI-powered app segmentation limits the blast radius of a potential incident.
  • Shut down compromised users and insider threats: Inline inspection prevents private application exploit attempts, and integrated deception capabilities detect the most sophisticated attackers.
  • Stop data loss: Inspection of data in motion and at rest prevents potential theft by an active attacker.

• Foundational security best practices: Learn fundamental security best practices to enhance overall resilience to phishing attacks.

Download your copy of the Zscaler ThreatLabz 2024 Phishing Report today. Phishing attacks will persist and remain a pervasive threat to organizations. By understanding the latest phishing trends, assessing the associated risks, and recognizing the implications of AI-driven attacks, your organization will be better equipped to defend against phishing in 2024 and beyond.

Virtual private networks (VPNs) have long provided remote access for enterprises. However, after a year full of high-profile exploits of severe and zero-day VPN vulnerabilities, including ones requiring emergency CISA directives to disconnect VPN services, enterprises are reconsidering. Many are rethinking workforce connectivity strategies and moving to zero trust architecture as a secure VPN alternative.

In fact, 91% of enterprises are concerned that VPNs will compromise their security.

Released today, the Zscaler ThreatLabz 2024 VPN Risk Report sheds light on critical VPN trends and explores solutions to secure remote users. Cybersecurity Insiders and Zscaler surveyed 647 IT professionals and cybersecurity experts on the shifting security, management, and user experience challenges of VPN technologies. These include the risks VPNs may present to an organization’s security posture around lateral movement, third-party access, and vulnerability to attacks like ransomware.

Overall, the findings paint a clear picture. Enterprises show stark concerns around the security risks of VPN solutions, even as they reach a strong consensus around zero trust strategies and make concrete plans to adopt zero trust network access (ZTNA). Organizations also agree hosting VPN infrastructure in the cloud is against the principles of zero trust and does not provide the same level of security or user experience as ZTNA.

Download the Zscaler ThreatLabz 2024 VPN Risk Report to uncover enterprise perspectives and expert guidance around the risks of VPN.

Key findings

• VPN attacks are on the rise. 56% of organizations experienced one or more VPN-related cyberattacks in the last year—up from 45% the year before. This highlights the growing frequency and sophistication of attacks targeting VPNs.
• The vast majority are shifting to zero trust. 78% of organizations plan to implement zero trust strategies in the next 12 months. Meanwhile, 62% of enterprises agree that VPNs are anti-zero trust.
• Most have doubts about VPN security. 91% of respondents expressed concerns about VPNs compromising their IT security environment. Recent breaches illustrate the risks of maintaining outdated or unpatched VPN infrastructures.
• VPNs are no match for ransomware, malware, and DDoS. Respondents identified ransomware (42%), malware (35%), and DDoS attacks (30%) as the top threats exploiting VPN vulnerabilities, underscoring the breadth of risks organizations face due to inherent weaknesses in traditional VPN architectures.
• The risk of lateral movement can’t be ignored. 53% of enterprises breached via VPN vulnerabilities say threat actors moved laterally, demonstrating containment failures at the initial point of compromise that underscores the risks of traditional, flat networks.
• Almost everyone has concerns about third-party risk. Since VPNs provide full network access, 92% of respondents are concerned about third parties with VPN access serving as potential backdoors into their networks.

Rising VPN attacks, CVEs, and enterprise concerns

Overall, a staggering 56% of organizations reported cyberattacks that exploited VPN vulnerabilities within the past year, marking a significant increase from the previous year (45%). Even more concerning, 41% of organizations reported experiencing two or more VPN-related attacks, highlighting the existence of severe security gaps that need immediate attention.

Figure 1: Enterprises that have experienced an attack that targeted VPN vulnerabilities in the past year.

This rise in VPN-related attacks is not without context. In the past year, we’ve seen a string of zero-day and high-severity VPN vulnerabilities come to light. This trend has revealed that, from an architectural point of view, VPN-based networks are vulnerable to a single point of failure that allows threat actors to move laterally on the network, discover crown jewel applications, and steal sensitive data.

Indeed, most survey respondents who experienced VPN-related breaches reported that attackers moved laterally on their networks.

Figure 2: A string of high-profile CVEs impacting VPN in the last year.

Enterprise trust in the security of VPNs is low. Overall, 91% of companies have concerns that VPNs may jeopardize ‌the security of their environments.

Figure 3: Enterprise concerns that VPN may jeopardize the security of their environment.

Growth in Zero Trust adoption

In parallel with, or because of, the security concerns of VPN, enterprises show a strong consensus around the adoption of zero trust strategies for secure connectivity. In fact, 62% of enterprises see VPN technology as incompatible with zero trust strategies.

Figure 4: Enterprise views on VPN as compatible with zero trust strategies.

Meanwhile, enterprises are actively adopting zero trust strategies as enthusiasm for VPN wanes. On the whole, 78% of enterprises plan to implement zero trust strategies within the next 12 months, while 31% are actively implementing zero trust strategies today.

Figure 5: Enterprise adoption of zero trust strategies.

As the number of high-profile security vulnerabilities associated with VPNs continues to rise, businesses should anticipate a corresponding rise in security incidents related to VPNs. As a result, enterprises will increasingly look to ZTNA as a replacement for VPN and a way to fundamentally improve their security posture.

Our ZTNA solution, Zscaler Private Access (ZPA) provides comprehensive security for users connecting to private applications regardless of their location, from any device. With ZPA, applications are hidden from internet exposure, making it difficult for attackers to find and target them. Our inline traffic inspection detects malicious activities to prevent compromise and data exfiltration. ZPA is able to limit ‌the blast radius with AI-powered user-to-app segmentation and integrated deception.

For the full report insights, including best-practice guidance on mitigating VPN risks and predictions for 2024 and beyond, download your copy of the Zscaler ThreatLabz 2024 VPN Risk Report with Cybersecurity Insiders today.

IT infrastructure provider Kyndryl has pumped up its portfolio of generative AI services by announcing a partnership with Nvidia to simplify genAI deployments for network and application management. It’s also offering new AI-driven mainframe modernization consulting and managed delivery offerings.

In terms of collaborating with Nvidia, Kyndryl said it will incorporate Nvidia NIM inference microservices into its Kyndryl Bridge integration platform to enable AIOps services to be optimized on Nvidia Tensor Core GPUs; the combination of technologies is aimed at allowing enterprises to quickly process failure prediction and analysis and significantly reduce network and IT infrastructure failures.

Kyndryl Bridge integrates disparate management, observability and automation tools while making use of AI and ML to analyze the aggregated data and provide IT operations teams with the intelligence they need to keep systems running at peak performance. NIM is a collection of AI-based microservices that can be deployed on any cloud or in a data center to help enterprises deploy generative AI across a variety of models and infrastructures, according to Nvidia. 

In addition, Kyndryl will incorporate retrieval-augmented generation (RAG) with Nvidia NeMo Retriever microservices to allow enterprise customers to tailor the package for specific management tasks, according to Kyndryl. The NeMo cloud-native platform lets users build, customize, and deploy generative AI models, and it includes training and inferencing frameworks, data curation tools, and pretrained models.

The vendors envision a number of use cases for the collaborative AI effort, such as establishing proper workload placement, operations automation, fraud and loss prevention, and real-time analytics.

“By combining Nvidia’s generative AI software with Kyndryl’s capabilities, we’re uniquely prepared to help address and resolve the biggest pain points for customers seeking to integrate AI across their hybrid IT estates,” said Hidayatullah Shaikh, vice president, software engineering, Kyndryl Bridge, in a statement.

In addition to the Bridge integration, Kyndryl Consult will support customer operations running on the Nvidia infrastructure, in on-premises deployments, or in a private cloud, hybrid or multicloud environments, the companies stated.

Using AI to transform mainframe environments

While GPU operations are the focus of the Nvidia collaboration, Kyndryl also said it will expand its AI-driven mainframe modernization services. The new AI services are aimed at furthering Kyndryl’s strategy of offering customers options when it comes to mainframe modernization – whether they want to keep data on the mainframe, migrate it off, or create a hybrid environment.

Kyndryl will now offer Bridge-based AI and generative AI tools for moving workloads off the mainframe to the cloud via generative AI-produced and Kyndryl-enhanced application documentation. The services will support automated conversion of classic mainframe application code to modern languages such Java.

Additionally, new AI-based services will help secure access to mainframe data used in cloud-based AI offerings, enable interoperability between mainframe and cloud operations, and assist in determining proper workload placement, Kyndryl stated.

“The AI-driven operational insights can enable more proactive and predictive management of mainframe systems, and provide visibility and control over mainframe performance and costs,” wrote Petra Goude, global practice leader for core enterprise & zCloud at Kyndryl, in a blog about AI and the mainframe. 

If customers are keeping data on the Big Iron, the AI tools can determine application code modernization techniques to avoid potential production issues. The tools will integrate real-time AI into existing mainframe applications for better business insights, according to Kyndryl.

“The convergence of AI, mainframe and cloud is shaping the evolution of IT. The world’s biggest financial institutions, manufacturers and healthcare providers have relied on the mainframe and classic programming languages like COBOL, PL/I and REXX since the beginning of enterprise computing,” Goude wrote. “The mainframe continues to serve these companies well. But the very nature of the mainframe – what enables its reliability, resilience and security capabilities – contributes to difficulties with systems modernization. This is where AI and generative AI can help.

Embedding AI into mainframe and hybrid cloud environments can help augment human capabilities, streamline business process automation, and generate actionable insights from data, Goude stated.

“In addition, organizations can optimize services delivery and hardware and software costs by implementing and deploying AI-enabled chatbots and other operational processes to help execute day-to-day operations and recommend technology best practices,” Goude wrote. “And running AI models on the mainframe can provide insights that can help companies enhance customer satisfaction and compliance with regulatory requirements and potentially reduce fraud losses.”