Home Blog

Google Cloud outage disrupts over 50 providers globally for over 7 hours



“Resilience isn’t a function you layer on. It’s an architectural dedication. Efficiency beneath adversity — not in good situations — is the actual benchmark now. In case your system can’t soak up failure with out taking your clients down with it, you’re not production-ready in 2025 — particularly not within the AI period,” stated Spencer Kimball, CEO & Co-Founder, Cockroach Labs.

Fragile dependencies throughout the web infrastructure

A world outage that originated inside Google Cloud’s inside infrastructure paralyzed providers throughout main platforms, highlighting a systemic vulnerability that underpin hyperscaler ecosystems.

“The domino impact from Google Cloud’s inside IAM failure was felt throughout dependent platforms like Cloudflare, Spotify, Snapchat, and Discord—not attributable to {hardware} failure, however as a result of control-plane dependencies paralyzed core administrative features,” stated Sanchit Vir Gogia, chief analyst and CEO at Greyhound Analysis.

Cloudflare acknowledged struggling a big service outage that affected a big set of its crucial providers, together with Employees KV, WARP, Entry, Gateway, Pictures, Stream, Employees AI, Turnstile and Challenges, AutoRAG, Zaraz, and elements of the Cloudflare Dashboard.

Based on the corporate weblog put up, the outage lasted 2 hours and 28 minutes and impacted all Cloudflare clients utilizing the affected providers globally. As part of infrastructure utilized by Cloudflare Employees KV service is backed by a third-party cloud supplier, which skilled an outage that immediately impacted the provision of the KV service, the corporate stated.

“This incident underscores the deep interdependence of at the moment’s web infrastructure. Whereas cloud suppliers seem technically unbiased, they typically share crucial components comparable to routing protocols, DNS providers, and edge supply techniques. These shared elements create systemic dangers, the place a failure in a single space can ripple throughout a number of platforms,” stated Manish Rawat, analyst, TechInsights. The occasion exposes the fragility of cloud redundancy, notably when core web protocols endure misconfigurations or outages.

AMD steps up AI competitors with Intuition MI350 chips, rack-scale platform



Different bulletins included ROCm 7, the newest model of AMD’s open-source AI software program stack, and the broad availability of its Developer Cloud, a completely managed platform geared toward accelerating high-performance AI improvement.

Openness and Nvidia problem

AMD underscored its dedication to open requirements and ecosystem collaboration, positioning itself in distinction to rival Nvidia, which relies upon closely on a proprietary software program stack.

“We’re coming into the following section of AI, pushed by open requirements, shared innovation, and AMD’s increasing management throughout a broad ecosystem of {hardware} and software program companions who’re collaborating to outline the way forward for AI,” AMD’s chair and CEO, Lisa Su, mentioned within the assertion.

The announcement follows AMD’s current acquisition of AI software program startup Brium, a deal the corporate mentioned introduced in deep experience to speed up the open-source instruments that energy its AI software program stack.

“Whenever you take a look at the specs, AMD’s MI355X is taped out on TSMC’s N3P course of, whereas Nvidia’s GB300 makes use of 4NP,” mentioned Neil Shah, vp for analysis and accomplice at Counterpoint Analysis. “This provides AMD a course of node benefit in efficiency and energy effectivity, particularly in comparison with the Nvidia Blackwell GB200/B200.”

Nonetheless, elements equivalent to reminiscence bandwidth, precision optimization, and software program framework help finally decide coaching and inference efficiency.

“AMD excels at optimizations for greater precisions (FP64, FP32) the place it holds a bonus, however Nvidia is healthier optimized for decrease precisions (FP4, FP8),” Shah added. “AMD now matches Nvidia on core capabilities, permitting it to compete head-on, which may result in improved theoretical TCO.”

Nonetheless, AMD’s ROCm software program stack continues to be catching as much as Nvidia’s extra established CUDA ecosystem, which stays a key think about real-world efficiency and complete price of possession. Broader adoption will depend upon how shortly AMD can slender this hole, although it continues to make regular progress with every era.

OpenAI o3-pro accessible within the API, BrowserStack provides Playwright assist for actual iOS gadgets, and extra – Every day Information Digest


OpenAI has introduced that o3-pro can now be accessed via the OpenAI API. Moreover, Professional and Group customers can now use o3-pro in ChatGPT.

The pricing for utilizing o3-pro within the API will likely be 87% cheaper than o1-pro, and the worth of o3 can be being minimize by 80%. “We optimized our inference stack that serves o3. Identical precise mannequin—simply cheaper,” the corporate wrote in a submit on X

BrowserStack provides assist for Playwright exams on actual iOS gadgets with Safari

In accordance with the corporate, it’s the first testing supplier to supply this. Beforehand, testers would wish to make use of Playwright’s desktop browser emulation to take action.

With BrowserStack’s newest assist, testers can run parallel exams throughout over a thousand actual iOS and Android device-browser mixtures, simulate real-world circumstances, and seize community, video, and textual content logs in a single place. 

“With iOS driving 28% of world internet visitors, testing Safari on actual iOS gadgets was the largest lacking piece in Playwright,” stated Nakul Aggarwal, CTO and co-founder of BrowserStack. “Now, groups can confidently run Playwright exams on actual iOS gadgets, guaranteeing constant, dependable outcomes throughout totally different browsers and gadgets.”

SmartBear launches Mirror Cellular

Mirror Cellular is a no-code cellular testing platform that helps testing on each iOS and Android. 

“Cellular take a look at automation stays complicated and underserved for a lot of QA organizations, even because it turns into a quickly rising requirement,” stated Prashant Mohan, VP of product administration at SmartBear. “Mirror Cellular, a significant step in advancing the SmartBear Check Hub technique, which brings API, internet, and cellular testing into one unified, clever resolution, was constructed to assist groups simplify cellular take a look at automation. We’re dedicated to serving to groups scale their software program high quality efforts with out growing complexity throughout instruments and workflows.”

Java by no means goes out of favor: Celebrating 30 years of the language


Regardless of having simply celebrated its 30 12 months anniversary, Java remains to be going sturdy. Scroll by means of the record of different programming languages created within the 90s and also you received’t discover many names you’d nonetheless acknowledge at present, but Java remains to be prevalent throughout the trade. 

TIOBE’s Programming Neighborhood index for June 2025 ranks Java because the fourth hottest programming language, Stack Overflow’s 2024 developer survey ranked it as seventh hottest, and IEEE Spectrum’s annual rating in 2024 positioned it within the quantity two spot. 

Growth on Java first started in 1991 by James Gosling, Mike Sheridan, and Patrick Naughton, and its first public launch was on Could 23, 1995. Through the first 15 years of the language, it was maintained by Solar Microsystems, which was acquired by Oracle in 2010, the place it lives at present. 

To know how Java has maintained its recognition over the many years, Simon Ritter, CTO of the Java platform Azul, says you must return to the philosophy Gosling had when the language was created. “He described it as a blue collar programming language, and what he meant by that was it was all about getting the job executed,” mentioned Ritter. “It wasn’t an educational train when it comes to investigating how we may change programming languages specifically methods and use various things. It was about, I must get a job executed. I need to get it executed in probably the most environment friendly approach. And that’s the way it happened, and that’s actually the philosophy that has adopted the language by means of these 30 years, and has had a big effect on why the language has continued to be so profitable.”

One more reason why the language has been so profitable is that it continues to tug options from different languages to maintain it related, Ritter defined. For instance, Textual content Blocks was a characteristic borrowed from JavaScript, purposeful programming was borrowed from Scala, and sample matching is one thing that has been seen in lots of different programming languages. 

“Different languages have come round they usually’ve added new issues, and folks have gone, ‘oh, I like that shiny new characteristic that’s been added to that language.’ And what Java has executed is claimed, ‘hmm, okay, let’s take a few of these concepts and produce these into the language,’” mentioned Ritter. 

And naturally, the “write as soon as, run wherever” nature of Java has been a draw as properly, particularly as {hardware} advanced. Brian Fox, co-founder and CTO of Sonatype, the corporate behind the Maven Central repository, mentioned “if you happen to go all the way in which again and also you have a look at C and C++, we’ve libraries and DLLs, however largely these issues needed to be recompiled on a regular basis for brand new variations of {hardware} and new chips and issues like that. It created this case the place to ensure that any of those blocks to be reusable, any person needed to continually replace them.”

The Java Digital Machine (JVM) eradicated that challenge and enabled a state of affairs the place you could possibly create the Java JAR file as soon as and by no means have to return to replace it. “You might focus your vitality on innovating that factor ahead versus simply continually rebuilding it to make it related within the trendy time,” he mentioned.

Ritter believes that the best legacy of Java is that nearly any web site that you simply work together with most likely has Java working someplace, particularly on the server aspect. 

“The JVM is admittedly the facility behind that and its potential to scale to tons of, 1000’s, thousands and thousands of customers, and nonetheless ship the issues that individuals want,” he mentioned. 

OpenJDK

A big milestone in Java’s historical past was the push to make the language extra open. 

“If we go proper again to the early days of Java’s historical past, it turned extremely popular, after which there have been quite a lot of calls from individuals to make it extra open, and Solar continued to keep up fairly strict management over it, in that they didn’t hand over the specification to an open requirements physique,” Ritter mentioned.

Nevertheless, Solar did acknowledge that there was a necessity for individuals to offer enter on how the language ought to evolve, which led to the creation of the Java Neighborhood Course of. 

A number of years later, in 2007, Solar launched the supply code of Java underneath the GNU GPL, resulting in the creation of OpenJDK. 

Switching the discharge schedule allowed the language to enhance extra regularly

In 2017, after the discharge of Java 9, Oracle switched from releasing a brand new model of Java roughly each three years to releasing two new variations yearly. This, Ritter says, has allowed the language to enhance fairly quickly ever since, whereas nonetheless bettering in a managed approach so that individuals don’t have radical modifications. 

In line with Fox, there was an inflection level between the discharge of Java 8 and 9, which is after they launched the module system and another revolutionary modifications. 

“It created just a little little bit of a fraction in the neighborhood we nonetheless see, the place individuals are struggling to replace their purposes previous Java 8 due to the beautiful vital modifications there,” he mentioned. 

He remembers working with Oracle and pleading them to not do issues that will create a schism in the neighborhood, like with Python 2 vs 3, and he believes they had been principally profitable at avoiding that. 

“There was just a little little bit of a hurdle to recover from it, undoubtedly not like a significant fork in the neighborhood the place each library needed to be recompiled within the new factor, otherwise you couldn’t use it,” mentioned Fox. “Apart from that, all the things feels largely incremental. I don’t assume there was another large watershed second, and I might assert that’s a very good factor. That’s why it continues to develop.”

Java EE turns into Jakarta EE

One other main change that occurred in 2017 was that Oracle donated Java EE—a set of specs for enterprise Java options—to the Eclipse Basis, the place it was renamed Jakarta EE. 

“From our perspective, Java is continuous to evolve proper earlier than our eyes, and it’s within the midst of a renaissance of kinds,” mentioned Tanja Obradovic, senior director of Java Applications for the Eclipse Basis. “Our group isn’t simply sustaining Java; it’s actively modernizing it, and builders proceed to show to it for all the things from cloud-native providers to AI.”

In line with Obradovic, Jakarta EE is an instance of how Java has advanced to help open-source builders and their wants. The subsequent launch will embrace deeper cloud native integration refinements whereas persevering with to deprecate outdated applied sciences. This helps to refine the platform to help applied sciences like microservices and event-driven architectures. 

“All of that is attainable because of the dedication of the open supply group. That presence is what has contributed to and even prolonged its longevity as a language,” she mentioned. 

Apache Maven

Apache Maven is a construct automation and dependency administration instrument for Java, and Fox mentioned it’s vital in Java’s historical past as a result of it made it very easy to construct software program with Java. 

“It was at all times deeply irritating to me as an skilled Java programmer, how arduous it was to attempt to make modifications to issues that we had been utilizing day by day. Like, our IDE was Eclipse, and it was written in Java, and it’s like, I actually want it might do that factor, and I ought to have the ability to take 20 minutes and simply change that code, however it was not possible to get it to construct. After which when Maven got here alongside, it was like, oh yeah, it’s tremendous straightforward to construct this code,” he mentioned.

Different languages which have adopted have adopted comparable methods for dependency administration, just like the npm package deal supervisor for JavaScript, pip for Python, and Nuget for .NET. 

“I really feel like these early issues that we did with Java actually helped open supply, not simply inside Java, however made all the things that got here after it have the identical primary form that enabled the large explosion that we’re seeing now,” mentioned Fox.

Cisco underscores AI dedication with networking LLM, agentic AI interface



“Workforce members can share dashboards with colleagues and save them to persist throughout periods,” Dhingra wrote in a weblog put up. “AI Canvas makes use of superior reasoning fashions to interrupt down troubleshooting into structured steps. When groups agree on a proposed resolution, AI Canvas takes motion to repair the issue, from implementing a configuration change to executing steps in a runbook. This collaborative strategy makes use of AI to deliver NetOps, SecOps, and different groups collectively for sooner downside decision.”

It’s a distinction maker, based on Neil Anderson, vp of cloud, infra, and AI options with IT service and world methods integrator World Vast Expertise. 

“Lastly, with the ability to work throughout all these completely different silos that an enterprise has needed to work with individually for years, and simply mitigate community and safety issues, is sport altering for patrons,” Anderson stated. As well as, because the Deep Community Mannequin is constructed on Cisco’s personal data base, the main points it feeds AI Canvas shall be far more correct than the types of information accessible from public methods, Anderson stated.

The Deep Community Mannequin additionally feeds the brand new Cisco Stay Shield system, which is a real-time, stay patching system that applies safety controls on the kernel stage to defend in opposition to newly noticed vulnerabilities and exposures with out requiring swap or router reboots. The Stay Shield characteristic might be triggered or managed by way of Cisco’s AI assistants or Hypershield workflows.

Agentic AI

Cisco talked extensively on the occasion in regards to the rising world of agentic AI, the place AI brokers can share info, collaborate, cause, and take autonomous actions in real-time, Sampath stated.

“Cisco’s imaginative and prescient integrates cutting-edge capabilities, together with automated agent discovery, delegated authorization, safe zero-trust agentic entry, and native assist for the Mannequin Context Protocol (MCP),” stated Jeetu Patel, Cisco’s president and chief product officer (pictured above at Cisco Stay keynote). MCP provides a standardized approach for AI fashions to work together with exterior instruments and companies, akin to code repositories, databases and net companies.