Presented by F5

Enterprise AI teams have spent years solving for compute, securing GPU allocations, negotiating cloud capacity, and benchmarking training throughput. The assumption embedded in that work is that the path between storage and compute will keep up. In production, that assumption increasingly does not hold. Real traffic introduces latency spikes, network jitter, and node degradation that controlled benchmarks fail to capture, resulting in pipelines that perform well in the lab but stall in deployment. A growing response is AI data delivery, deploying an application delivery controller (ADC) or application delivery and security platform (ADSP) in front of storage as a resilient and secure control point.

“Provisioning solves for capacity but not for delivery, and that is where the constraint now hides,” says Hunter Smit, senior manager of product marketing at F5. “Enterprises buy enough GPUs and enough storage, then assume the path between them will keep up, but AI traffic is bursty, highly concurrent, and random in its reads in ways ordinary storage networking was never built to absorb.”

The production gap benchmarks don’t show

Standard benchmark methodology compounds the problem, says Paul Pindell, principal solutions architect for technology alliances at F5.

“Benchmark testing is usually built to produce the best possible performance or security result, not the most realistic one,” he says. “With S3, latency is a known factor in degrading performance, so meaningful testing has to introduce consistent latency into the path.”

Most benchmark environments never do that, which means the performance numbers enterprises rely on for infrastructure decisions are drawn from conditions that production systems will never replicate. To test this assumption, F5 and MinIO conducted throughput testing under degraded network conditions.

“What stood out was how quickly S3 throughput falls off once you introduce latency,” Pindell says. “Even modest latency takes a real bite out of it, and as latency climbs toward long-haul distances, the degradation gets severe.”

The testing also showed latency mattered far more than jitter as a driver of throughput loss, which inverted what the team had expected going in. The upshot for enterprise architects is that S3 object storage deployments cannot be designed around clean-room assumptions; they have to be engineered for the degraded network conditions they will actually face.

The cost of fragile data paths

“In AI infrastructure, people naturally focus on GPUs because they’re the most visible and expensive resource,” says Tanu Mutreja, senior director of product management at F5. “But in production environments, GPUs generate only as much value as the data path that feeds them.”

That path runs through storage, networking, databases, security, and orchestration layers, often stitched together from multiple vendors. Customers experience none of those seams; they experience the output of the whole system.

When the data path degrades, the effects compound. GPU underutilization is the most immediate and visible symptom, but Mutreja pointed to a wider set of consequences: degraded inference performance, poor-quality AI outputs, higher egress costs from unnecessary data replication, and growing operational complexity.

“At scale, data-path efficiency becomes a strategic business lever rather than technical optimization,” she says. “When the data path is engineered well, GPUs remain productive, AI applications stay responsive and trustworthy, operations scale efficiently, and organizations maximize the return on their AI investments.”

AI workloads are structurally more exposed to these failures than traditional enterprise applications. Databases, ERP systems, and web services absorb transient storage delays through caching and buffering. AI workloads running across massively parallel GPU clusters have no equivalent protection. As Mutreja noted, even minor latency spikes or bandwidth bottlenecks can cascade across large GPU clusters, simultaneously hitting utilization, training efficiency, and the customer experience.

Treating the storage edge as a control point

For decades, storage and intelligence operated as sequential concerns in enterprise architecture: data was stored first, then analyzed downstream. Mutreja argued that this model no longer fits the demands of AI.

“Competitive advantage is determined not only by the volume of data, but also by relevance, lineage, security, and performant delivery of data,” she says. “Across the industry, from NVIDIA and AWS to enterprise storage providers, the movement is toward embedding intelligence directly into data infrastructure rather than stacking it on top.”

F5’s integration with MinIO instantiates this approach at the layer where storage and compute actually interact. As part of the F5 ADSP, BIG-IP sits in the data path, continuously monitoring the health of MinIO’s distributed storage nodes and directing requests only to those that remain available.

The operational impact of that capability becomes clear when nodes degrade, which is expected in distributed storage clusters. Without intelligent routing, clients that land on an unhealthy node must retry and may land on another degraded node, dragging down overall performance.

“F5 makes sure traffic only goes to healthy nodes, or even the least busy ones, so S3 client traffic is always processed in the most efficient way,” Pindell says.

Governance across distributed environments

The challenge grows at scale, when AI pipelines stretch across multiple locations, clouds, or edge environments.

“Once an AI pipeline crosses regions and clouds, the question stops being about performance and becomes about control,” Smit says. “You are operating under different rules in every jurisdiction, and digital sovereignty is now a design constraint. Where your data is allowed to live, who is permitted to touch it, and which borders it cannot cross now shapes the architecture before anyone talks about speed.”

That pressure is driving a visible trend of enterprises repatriating AI workloads from public cloud onto infrastructure they own and govern directly. The architecture Smit described resolves this by decoupling applications from any single storage location and placing a unified control point between them that enforces consistent policy across all of them.

“Sovereignty, resilience, and cost stop being trade-offs you manage one region at a time,” he explains. “They become a capability you run as a system.”

Storage-to-compute path as a managed control point

To solve for these issues, enterprise teams need to stop treating the storage-to-compute path as a direct connection and start treating it as a managed control point, Smit says. SecureIQLab’s independent validation of F5 BIG-IP in storage deployments has confirmed the approach delivers resilience without surrendering throughput.

“Insert a full-proxy ADC between the two, and the path becomes observable, programmable, and failure-aware, with health-based routing, quality of service, and security enforced inline,” he explains. “That single move converts data delivery from an assumption into an engineered discipline, which is what keeps GPUs fed when conditions degrade.”

Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.

It’s early days yet, but if you rely on vendors’ software for your RAID enclosure, you probably need to find out what their macOS 27 plans are.

We’ve had macOS 27 for all of four days at this point, and there may already be a show-stopping problem for folks that hang on to RAID enclosures. We’ve found several that just don’t work under macOS 27.

For example, I’ve got a Thunderbolt 3 LaCie 12Big enclosure that I’ve had for a while. It runs fine in Tahoe, on a Mac mini home server that I’ve had for years.

That LaCie 12Big doesn’t work at all in macOS 27. I’ve tried fresh installs of the software, different cabling, updating a Mac in place that it worked on in macOS 26, nothing works.

All of my dumb enclosures work and mount fine. SoftRAID as it stands now from OWC works fine in macOS 27 to set up a new array on those dumb enclosures.

This has happened before, of course. Talk to Drobo and Pegasus enclosure owners about when Apple changed how it handled device drivers a few years ago. It’s just worth mentioning that it’s happening again.

To get in front of this, I’m not talking about RAID arrays with DIP switches, other physical ways to configure the drives, arrays set up with Disk Utility, or Network Attached Storage devices. This is about vendors that sell enclosures that need special software to run on macOS.

Who’s at fault, and why is this happening?

I wish I had a good answer for you. Apple does like changing things, like how it’s done something with the boot selector in macOS 27. So there’s something there.

Also, macOS 27 also ships with no Intel code remaining, which could affect drivers compiled for Intel-only targets.. That may have something to do with it too.

Beta cycles are intended for developers to update their software to the new macOS. They exist for testing things like this.

But in our experience in the past, some things made by third party vendors get left behind.

Our advice as always stands. If you have mission critical hardware, this is not the time to try out the betas. And, if you’ve got enclosures that rely on older software, like my LaCie 12Big, it’s time to contact the vendor to see what’s going on.

And come up with a plan if there won’t be support in macOS 27.

It’s the latest case to raise alarms about ChatGPT’s lack of safeguards for suicidal behavior.

from the vampire-squid-strikes-again dept

To lose one speech-suppressing SLAPP suit may be regarded as thoughtless. To lose two looks like you’re a censorial hack.

Last month we wrote about how supposed “free speech warrior” Matt Taibbi (who spent years misrepresenting the work of people who study disinformation as inherently censorial, while getting pretty basic facts wrong) had lost his speech suppressing SLAPP suit against author Eoin Higgins. In that case, he argued that some rhetorically hyperbolic metaphors used on the book’s cover defamed him. The court pointed out that’s not at all how defamation works.

Taibbi, who also claimed he somehow had to sue to “protect free speech” (also not how it works) apparently wasn’t satisfied with just a single SLAPP suit. He also had sued congressional Rep. Sydney Kamlager-Dove in a separate action, claiming that her calling him a “serial sexual harasser” (and entering into the record two articles to support that claim) during a congressional hearing was defamation. If you’re interested, the two articles that were entered into the record were the Chicago Reader’s “Twenty years ago, in Moscow, Matt Taibbi was a misogynist asshole—and possibly worse” and the Washington Post’s “The two expat bros who terrorized women correspondents in Moscow.“

The hearing in question was yet another in a ridiculously long line of congressional hearings (multiple ones where Taibbi has appeared peddling nonsense) about the supposed “censorship industrial complex,” a mostly made-up concept pushed by political hacks trying to shield online trolls and bullies from ever facing consequences from private actors for breaking the clearly stated policies of online platforms.

Kamlager-Dove chose to question Taibbi’s credibility. You could argue she could have focused on the factual problems with his continued confused claims about how disinformation research and trust & safety work — but she went for the more salacious (and widely reported) claims about his time in Moscow from a few decades ago, along with a characterization that reads as a clear opinion based on disclosed facts, which (by definition) cannot be defamatory.

As you may be aware, things said in Congress tend to be protected by the speech and debate clause of the Constitution. Taibbi’s lawyers claimed that because Kamlager-Dove reposted videos of her remarks on social media, that somehow took them outside the clause’s protection. For her part, Kamlager-Dove pointed to the Westfall Act which (as we’ve discussed in the past) allows the government itself to substitute in as a defendant in cases filed against government employees if the lawsuit was based on government work they were doing. In defamation cases, this is fatal: once the federal government substitutes itself in as defendant, the case collapses, because you simply can’t sue the federal government for defamation thanks to sovereign immunity.

Here, the case fails on those grounds exactly. Judge Evelyn Padin finds that the Westfall Act does apply, effectively dooming the case. Taibbi’s lawyers tried to argue that Kamlager-Dove’s statements weren’t part of her job as Congress… because her comments were “partisan communications” and were for “self-aggrandizement on Twitter” rather than serving her constituents. Except politicians making self-aggrandizing partisan communications is (unfortunately) part of their job these days.

Representative Kamlager-Dove’s Statements and republications, however, are precisely the kind of conduct that is “a central part of the job for members of Congress.”…. Indeed, a “primary obligation of a [m]ember of Congress in a representative democracy is to serve and respond to his or her constituents.” …. As the Ranking Member of the Subcommittee holding the Hearing. Representative Kamlager-Dove’s remarks mentioned “taxpayer time and resources” and “foreign policy” topics that are important to members of Congress and that are top-of-mind for their constituents….

Republishing the statements online does not change the analysis. Taibbi claims that the “republications on X, BlueSky, and [Representative Kamlager-Dove’s] website were not legislative work, [and] occurred outside the legislative setting.” …. But members of Congress routinely engage with the public on social media and on the internet as part of their jobs…. (“There is no meaningful difference between tweets and the other kinds of public communications between an elected official and their constituents that have been held to be within the scope-of-employment under the Westfall Act.”). As Taibbi concedes, Representative Kamlager-Dove was simply “talking to voters on Twitter.” …

Thus, while the judge doesn’t get a chance to dismiss the censorial SLAPP suit for being a censorial SLAPP suit, the court does make it pretty clear you can’t sue over this kind of thing.

Two SLAPP suits filed to silence critics. Both dismissed. This is a guy who built his recent brand on the Twitter Files and the “censorship industrial complex” — and who has been a key cog in helping the government suppress speech in the process. He’s now spent quite a lot of time trying to use the courts to shut people up for criticizing him — and failing at that, too.

Filed Under: defamation, free speech, matt taibbi, slapp, slapp suit, sydney kamlager-dove, westfall act

• Hybrid meetings can leave remote workers feeling excluded, Jabra study finds
• Unsuitable and dated setups cause regular meeting delays and technical failures
• Better meeting room kit and clear meeting purposes could improve engagement

Around half of remote participants say they’re forgotten, talked over or excluded during hybrid meetings, a new study from Jabra has revealed, indicating that hybrid in-person and remote meetings might not be as effective as we’d thought.

The issue is particularly evident when multiple participants are in a physical room, with others joining online. But more than that, women (16%) and junior workers (26%) are more likely to feel they’re being excluded.

I am going to start where no good teacher should start, with a $10 word: epistemology. It refers to a branch of philosophy that explores how we know what we know – something scholars like John Dewey argued is deeply tied to experience, not just information.

This word takes me back to my doctoral graduation when my father-in-law said with good-natured humor, “Well, Ev… there’s a lot of [stuff] you can’t learn from a book.” At the time, I didn’t know what to say, but any teacher worth their salt will tell you: he’s right.

Pre-service teachers – myself included – often lament that they didn’t really learn to teach until the rubber-meets-the-road experience of student teaching or that first job. This is the challenge of teaching pre-service teachers. I’ve been doing it for a handful of years now, and I see a trend – the TikTok way of knowing in education. It’s got me wondering how we adapt our practices based on my experience during my recent final exams with pre-service teachers.

The TikTok way

For example, I ask my students to make two tangible items to try and circumvent AI. One item is a teacher creed. I hand out “fancy” paper and tell them to create something they might read every teaching day – something to remind them not if, but when teaching gets hard. These are heartfelt, colorful creations. They write things like, I will show up with a good attitude. Even on my worst day, I will be someone’s favorite teacher. I cringe a bit, knowing how more seasoned educators might scoff but that is perhaps why I assign them – to bottle that early hopefulness in a landscape that often doesn’t often create it for new teachers.

The second item is to create “One One-Pager to Rule Them All!” Students make non-linear, doodle-style notes throughout the semester, and this final asks them to zoom out and represent everything essential we’ve learned through a map of connections, images, and ideas.

I love this assignment because I can see who is connecting the dots and who is simply regurgitating the text. I sit with each student for five to seven minutes as they “show and tell” the work. As they read their creeds, I am heartened and sometimes even tear up. And in conversation after conversation this semester, I heard the same phrase, almost as a confession mid-conference:
  “I know it’s not research-y, but in a TikTok I saw…”
  “I know it’s not the best source, but I saw a reel that said…”
  “This guy I follow always says…”

Each of these notes expanded or connected my own thinking about course content. Some couldn’t be backed in my mind of research, but others could. So, instead of arguing, I asked questions: Who created that content? What might their motivation be? Why does it matter to you? This kind of questioning reflects what Marilyn Cochran-Smith and Susan Lytle describe as “inquiry as stance” – an orientation where teachers are active investigators of knowledge.

An epistemological shift

We are in a shift in epistemology. Future teachers are learning not only through peer-reviewed research or textbooks, but also through short-form video, personality-driven content, and lived teacher experience shared in real time – what media scholars like Henry Jenkins describe as a more participatory culture of knowledge. This is democratizing, the dismantling of the silo that has long held educational research out of reach. But this is also destabilizing.

During my first years of teaching, I cried in my car a lot. If I had had the megaphone of TikTok influencers celebrating how they left education, or even my own content microphone, I’m not sure I would have made it through to my later years of teaching that are still hard but more grounded and fulfilling.

Admittedly, some positions are ones to leave. Yes, at times educator working conditions are not what they should be but how do we help pre-service and early-career teachers move through the baptism-by-fire years while being bombarded by voices – many from people who have left the profession and now narrate it from the outside? Some of the content is helpful. Some of it is not. And all of it is loud.

I wonder if our teacher preparation programs are keeping pace with how knowledge is actually being formed. It leads me to my favorite teacher question, “So what? What do we do now?” How long do we hack away at the plant growing up the wall, and when is it time to embrace the aesthetic of a vine-covered building as something worth studying?

Instead, what if instead we become weavers of stories? What if we help students craft their own and build connections of knowing? What if we engage lived experience not as secondary to research, but as a complementary form of knowing? When have we had so much access to real-time teacher voices about things that happened to them in the classroom that day?

Just because something is visual, narrative, click-baity, and social doesn’t mean it is missing the mark or doesn’t engage a pedagogical question worth exploring. This TikTok wondering is happening whether we embrace it or not, so what if we see it as a new charge to help future teachers engage these voices critically, rather than pretending they don’t exist?

Here are some ideas I’m playing with. I’m curious what you might add.

Ed Content Fridays. Students bring in content that connects with the week’s readings and learning from their own scrolling. Discuss it in a Spider-Web format that employs elements of a librarian CRAAP test to help students develop habits of mind around credibility and content creator motivation.

Use a C3WP writing strategy that engages reels and posts to kick off class. Start with what students know as a free write and then bring in content to have them expand their arguments and defend thoughts with research from our shared text.  If students bring it in, they find it interesting, and we can require a citation connection to the course text or researchers.

Like/Share/Subscribe. Share strong online content that sings from reputable sources with students. Syllabi and course hubs can be places to curate rich content collaboratively.

Have students create their own content. CapCut on a desktop or Edits on a phone are surprisingly easy plug-and-play tools to make short form videos, and we can up the academic requirements with or without student posting. Thoughtful content can grow out of our rich history of educational research, bringing rich, thoughtful voices in among the pervasive ranting. I’m not saying we shouldn’t be about the work of educational reform and that a good rant doesn’t have its place, but this new way of knowing and sharing knowledge is sitting in our desks waiting for us to light the fire.

Yes, my step-dad is right, there is so much we can’t learn from a book, but maybe there is still so much we can learn from our own students in their own ways of knowing, even if we don’t fully understand them ourselves. What if our ways of knowing weave together, creating something beautiful?

Congress is reviving one of the most significant antitrust bills Apple has faced in years, reopening a fight over the App Store and platform control that the company helped spend millions to defeat during previous congressional sessions.

Sens. Amy Klobuchar, D-Minn., and Chuck Grassley, R-Iowa, reintroduced the American Innovation and Choice Online Act (AICOA) on June 10. It revives a bipartisan effort to limit how dominant technology companies favor their own products and services.

The bill targets the largest online platforms and seeks to restrict conduct that supporters say gives those companies an unfair advantage. Apple and other technology giants spent years fighting earlier versions of the legislation because of its potential impact on their businesses.

The proposal would prevent dominant technology companies from favoring their own products and services. Lawmakers describe those practices as self-preferencing and argue they can disadvantage competitors.

Critics argue Apple uses its position as the operator of iOS and the App Store to benefit its own services over competing products. The legislation could directly affect the App Store and Apple’s control over the iPhone ecosystem.

Apple has consistently argued that its policies help protect user privacy, security, and the integrity of its platforms. In a statement provided to AppleInsider, Apple said it “strongly disagree[s] with the Senate’s consideration of European-style regulation” and argued the legislation would undermine privacy, security, and child safety protections while making it harder to do business in the United States.

The company also said importing Europe’s “failed policies” would not increase competition. The reintroduction marks the latest chapter in a legislative battle that has stretched across multiple sessions of Congress.

Earlier versions of AICOA advanced through the Senate Judiciary Committee but never reached a final vote despite bipartisan support. The bill came closer to becoming law than many technology reform proposals.

The debate around AICOA has changed since Congress first considered the legislation. Apple has already made significant App Store changes in Europe to comply with the Digital Markets Act.

The European law imposed new requirements on how large technology platforms compete and operate. The DMA and AICOA take different approaches to regulation.

Both aim to limit how dominant technology companies use control of their platforms to benefit their own products and services. For Apple, the DMA offers a real-world example of the kinds of changes lawmakers have sought through AICOA.

The company argues AICOA would mirror key elements of Europe’s Digital Markets Act, which required the company to make significant App Store changes in the European Union. According to Apple, the DMA has weakened privacy protections, increased security risks, and created a more difficult environment for product launches and platform development.

Why Apple fought the bill

Apple was among several technology companies that opposed the legislation during its previous runs through Congress. It argued that some provisions could make it harder to maintain privacy and security protections on its platforms.

Industry groups representing large technology companies also warned that the legislation could have unintended consequences for integrated products and services.

Supporters argue dominant platforms have too much control over businesses that depend on them. They say existing antitrust laws haven’t done enough to address those concerns.

Major technology companies spent heavily to stop AICOA and related antitrust legislation. Previous reporting found that Apple, Amazon, Google, and Meta collectively spent more than $100 million on lobbying and advocacy efforts tied to the proposals.

Trade groups also joined the fight, and industry-backed advertising campaigns helped amplify the opposition. The legislation ultimately stalled despite advancing through committee and attracting support from both parties.

Why the legislation matters now

The bill’s return doesn’t guarantee it will become law. Previous versions generated substantial attention and bipartisan support but ultimately stalled before reaching the finish line.

For Apple, the debate extends beyond another round of regulatory scrutiny. The legislation could affect how the App Store operates and how Apple Services compete on the company’s platforms.

Whether the latest version gains enough support to advance remains unclear. Its return shows that Congress is still trying to limit how dominant technology platforms use control of their ecosystems to benefit their own products and services.