Artificial Intelligence

Juniper to partner with the 2026 Olympics and Paralympics

The Milano Cortina 2026 Foundation and Juniper Networks, a leader in secure and AI-Native Networking, have signed a partnership agreement for Milano Cortina 2026 Olympic and Paralympic Winter Games. The collaboration aims to optimise network systems and protect data and virtual information for the major sporting event. Making its Olympic and Paralympic debut as a partner of the Milano Cortina 2026 Olympic and Paralympic Winter Games, Juniper’s entry significantly enhances the project and the organising committee’s team. Juniper will help manage the event’s digital complexities such as real-time data management, cybersecurity and high-volume network demands, enabling a smoother operation. Sujai Hajela, Executive Vice President, AI-Driven Enterprise at Juniper Networks, comments, “Juniper leverages the right data, the right real-time response and the right infrastructure to provide predictable, reliable, measurable and secure connections for every device, user, application and resource. With our unique AI-Native Networking Platform featuring industry-leading wired, wireless, routing and security solutions, users will have simple and reliable access to digital assets and online information throughout the Milano Cortina 2026 Olympic and Paralympic Winter Games.” The partnership is built on technological excellence and shared values of inclusion and employee well-being, Juniper states. Furthermore, the company says that it prioritises people, equality and diversity - principles which align with the organising committee’s vision, making the Games a model of people-centric values.  Chris Barnard, Vice President, Telecoms and Infrastructure (Europe), IDC, adds, “The partnership between Milano Cortina 2026 Foundation and Juniper Networks highlights the experience-first value that intelligent technology can bring to large-scale sports events. Juniper’s AI-Native Networking Platform is designed to provide the data-driven reliability and security needed to manage the relentless demands of the Olympic and Paralympic Winter Games, delivering exceptional connectivity and robust protection of digital assets.” Mario Manfredoni, Senior Sales Director, South Europe, Juniper Networks, concludes, “Juniper’s partnership with Fondazione Milano Cortina 2026 will showcase Juniper’s services and high-performance networking. As the official secure IP network provider of the Milano Cortina 2026 Olympic and Paralympic Winter Games, Juniper will support in addressing the complexity of the major event across multiple locations. "Additionally, Juniper will employ a circular economy approach by collecting and re-cycling all equipment once the event concludes. Juniper’s goal is that none of the equipment goes to waste or ends up in landfill, and instead foster a more environmentally friendly business model. The products will be returned to the production process as pre-owned items through Juniper’s partner." For more from Juniper, click here.

NetApp receives AAA rating for its AI ransomware detection

NetApp, the intelligent data infrastructure company, today announced that NetApp ONTAP Autonomous Ransomware Protection with Artificial Intelligence (ARP/AI) has received a AAA rating from SE Labs, an independently-owned and run testing company that assesses security products and services. SE Labs validated the protection effectiveness of NetApp ARP/AI with 99% recall - a metric that measures malware detection rates - for ransomware attacks, while noting the absence of false positives. When responding to ransomware attacks, seconds can make the difference between ensuring continuity and a massive business disruption. Organisations need fast, automated, and accurate detection and response capabilities built into their primary storage systems to minimise the damage done by lost production data and downtime. NetApp ARP/AI, with its AI-powered ransomware detection capability, addresses this gap by providing real-time detection and response to minimise the impact of cybersecurity threats. SE Labs rigorously tested NetApp ARP/AI against hundreds of known ransomware variants with impressive results. NetApp ARP/AI demonstrated 99% detection of advanced ransomware attacks. NetApp ARP/AI also achieved 100% detection of legitimate files without flagging any false positives, indicating a strong ability to operate in a business context without contributing to alert fatigue. Ensuring data is secure against internal and external threats is a critical part of making data infrastructure intelligent, which then empowers customers to turn disruption into opportunity. This validation of NetApp’s AI-powered ransomware detection capabilities underscores how NetApp is staying at the forefront of AI innovation by both enabling AI adoption and applying AI to data services. NetApp’s newly released more powerful, all-flash storage systems help enterprises leverage their data to drive AI workloads, built on NetApp’s secure storage infrastructure. "NetApp has passed a significant milestone in the fight against ransomware as the first and only storage vendor to offer AI-driven on-box ransomware detection with externally validated top-notch protection effectiveness,” says Dr. Arun Gururajan, Vice President, Research & Data Science at NetApp. “Ransomware detection methodologies that rely only on backup data are too slow to effectively mitigate the risks businesses face from cybersecurity threats. NetApp ARP/AI hardens enterprise storage by providing robust, built-in detection capabilities that can respond to ransomware threats in real time. The AAA rating we achieved from SE Labs is the result of our commitment to innovation in intelligent data infrastructure and our drive to find new ways to make NetApp the most secure storage on the planet.” By embedding ransomware detection in storage, NetApp ARP/AI helps customers improve their cyber resilience while reducing the operational burden and skills required to maintain their intelligent data infrastructure. NetApp ARP/AI’s detection technology continuously adapts and evolves as new ransomware variants are discovered. This continuous retraining on the latest ransomware strains ensures that NetApp ARP/AI remains at the forefront of protection effectiveness, offering organisations a future-proof defence against the dynamic ransomware landscape. To see the full results of the tests, read the SE Labs Report by clicking here. NetApp ARP/AI is currently in tech preview. Customers can request to participate in the tech preview by reaching out to their NetApp sales representative. For more from NetApp, click here.

Pure Storage introduces unified data storage platform

Pure Storage, an IT pioneer that delivers advanced data storage technologies and services, has announced new capabilities in the Pure Storage platform that are evolving the ways IT and business leaders can improve their ability to deploy AI, improve cyber resilience, and modernise their applications. The Pure Storage platform delivers agility and risk reduction to organisations with a simple, consistent storage platform and an 'as-a-service' experience for the broadest set of use cases across on-premises, public cloud, and hosted environments. At the heart of the platform, the Evergreen architecture brings continuous and non-disruptive upgrades helping enterprises adapt to dynamic business environments. With the industry’s record number of concurrent SLAs, customers get the reliability, performance, and sustainability their business requires. Charles Giancarlo, Chairman and CEO, Pure Storage, says, “Pure is redefining enterprise storage with a single, unified data storage platform that can address virtually all enterprise storage needs including the most pressing challenges and opportunities IT leaders face today, like AI and cyber resilience. The Pure Storage platform delivers unparalleled consistency, resilience, and SLA-guaranteed data storage services, reducing costs and uncertainty in an increasingly complex business landscape.” Pure Storage announced new innovations in the platform including: Storage automation: Pure Fusion unifies arrays and optimises storage pools on the fly across structured and unstructured data, on-premises, and in the cloud. Now fully embedded into the Purity operating environment designed to continually get better over time via non-disruptive upgrades, the next generation Pure Fusion will be available across the entire Pure Storage platform to all global customers. Generative AI co-pilot for storage: Extending Pure Storage’s leadership position as the innovator in simplicity, the first AI co-pilot for storage represents a radically new way to manage and protect data using natural language. This leverages data insights from tens of thousands of Pure Storage customers to guide storage teams through every step of investigating complex performance and management issues and staying ahead of security incidents. In the new Innovation Race survey of 1,500 global CIOs and decision makers commissioned by Pure Storage, nearly all respondents (98%) state that their organisation’s data infrastructure must improve to support initiatives like AI - which is evolving so rapidly that IT is struggling to keep up, much less predict what’s next. Companies large and small are realising that they are locked into inflexible storage architectures lacking enterprise-grade reliability, unable to resize or upgrade performance without complex and risky infrastructure planning. Pure Storage is introducing new innovations in the platform that help businesses accelerate successful AI deployments today, and future-proof for tomorrow. The Pure Storage platform empowers organisations to unlock the value of their data with AI, while delivering agility to instantly scale capacity and performance up and down independently, without disruption. New Evergreen//One for AI - First Purpose-Built AI Storage as-a-Service: Provides guaranteed storage performance for GPUs to support training, inference, and HPC workloads, extending Pure Storage's leadership position for capacity subscriptions and introduces the ability to purchase based on dynamic performance and throughput needs. The new SLA uniquely delivers the performance needed and eliminates the need for planning or overbuying by paying for throughput performance. Secure Application Workspaces with Fine-Grained Access Controls: Combines Kubernetes container management, secure multi-tenancy, and policy governance tools to enable advanced data integrations between enterprise mission-critical data and AI clusters. This makes storage infrastructure transparent to application owners, who gain fully automated access to AI innovation without sacrificing security, independence, or control. For more from Pure Storage, click here.

Avaneidi secures funding to advance data security

Avaneidi, an innovative Italian start-up specialising in security enterprise storage systems, has announced an €8 million (£6.7m) Series A funding round by United Ventures. The investment underscores a shared commitment to advancing solid-state storage technologies, enhancing data security, and promoting a sustainable digital transition. Avaneidi develops comprehensive enterprise storage systems based on a rigorous 360-degree, multi-level 'security by design' approach, enabling an unprecedented degree of cyber security, protection and data reliability for enterprise-grade applications. Avaneidi’s storage technology advancements boost performance, security and reduce energy consumption. This allows electronic devices and data centres to increase their operating efficiency and limit their carbon footprint, addressing key sustainable development goals such as clean energy and sustainable industry innovation. Avaneidi’s Enterprise Solid State Drives (ESSDs) utilise tailor-made chips and advanced algorithms, providing a bespoke solution optimised for performance and cyber security applications. Designed for on-premise data centres, their storage appliances offer a cost-effective, highly efficient alternative to traditional storage solutions, featuring extended drive lifetime, improved security and significant energy savings. “Our mission at Avaneidi is to pave the way for more secure, efficient, and sustainable data storage solutions,” says Dr. Rino Micheloni, CEO of Avaneidi. “This funding will keep us at the forefront of the market, enabling us to accelerate the development of our enterprise ESSDs and all-inclusive storage appliances. Unlike off-the-shelf products, our solutions address cyber security and data governance issues by leveraging a tight hardware-software co-design while offering extensive customisation options.” Avaneidi targets organisations and industries that are highly sensitive to data governance and security, particularly within the rapidly evolving field of AI applications, where these issues are of paramount importance, such as finance, defence, automotive and healthcare. By prioritising data integrity and protection, Avaneidi empowers entire industries to better leverage AI technology safely and effectively when it comes to storage solutions. Avaneidi’s technology’s potential has attracted the attention of major industry players, the company states. Negotiations and preliminary agreements are in place to validate and expand the market reach of its innovative products. “United Ventures invests in technologies that have a tangible positive impact,” states Massimiliano Magrini, Managing Partner at United Ventures. “Avaneidi's vision and mission to enable organisations to make better and more sustainable storage decisions, focusing on governance and data security, align with our investment philosophy. By channeling resources into AI infrastructure like Avaneidi’s, we aim to facilitate the development of technologies that will redefine industries and transform tomorrow's society.” As the AI sector rapidly expands, robust infrastructure for advanced AI applications is paramount. According to recent estimates, the AI infrastructure market is projected to grow from $25.8 billion (£20.3bn) in 2022 to $195 billion (£153.9bn) by 2027, reflecting a compound annual growth rate (CAGR) of 50%. This surge is driven by significant advancements in AI computing, which is expected to escalate from $15.8 billion (£12.4bn) in 2022 to $165 billion (£130.2bn) in 2027, achieving a 60% CAGR.

Vertiv launches new AI hub

While artificial intelligence (AI) use cases are growing at an unprecedented rate, expert information is scarce for pioneering data centres. Vertiv, a global provider of critical digital infrastructure and continuity solutions, recognises this knowledge gap and the urgency of accessing this information, leading to the launch of its AI Hub. Partners, customers, and other website visitors will have access to expert information, reference designs and resources to successfully plan their AI-ready infrastructure. The Vertiv AI Hub features white papers, industry research, tools, and power and cooling portfolios for retrofit and greenfield applications. The new reference design library demonstrates scalable liquid cooling and power infrastructure to support current and future chip sets from 10-140kW per rack. Reflecting the rapid and continuous changes of the AI tech stack and the supporting infrastructure, the Vertiv AI Hub is a dynamic site that will be frequently updated with new content, including an AI Infrastructure certification program for Vertiv partners. “Vertiv has a history of sharing new to world technology and insights for the data centre industry,” says Vertiv CEO Giordano (Gio) Albertazzi. “We are committed to providing deep knowledge, the broadest portfolio, and expert guidance to enable our customers to be among the first to deploy energy-efficient AI power and cooling infrastructure for current and future deployments. Our close partnerships with leading chipmakers and innovative data centre operators make us uniquely qualified to help our customers and partners on their AI journey.” “AI is here to stay, and Vertiv is ready to help our customers navigate the challenges of realising their AI objectives. The AI hub is an excellent source for all our partners and customers in the Asia region to gain a deeper understanding of the opportunities and challenges of AI, and how Vertiv can assist to scale and embrace the AI journey,” says Alvin Cheang, High Density Compute Business and AI Director at Vertiv in Asia. Sean Graham, Research Director, Data Centres at IDC, notes, “Virtually every industry is exploring opportunities to drive business value through AI, but there are more questions than answers around how to deploy the infrastructure. A recognised infrastructure provider like Vertiv is valuable to businesses building an AI strategy and looking for a single source for information.” For more from Vertiv, click here.

SELECT warns about demand for electricity from power-hungry AI

SELECT's new President has warned that the demands on the electrical network to power AI may become unsustainable as it becomes a growing part of society. Mike Stark, who took over the association reins last week, said the UK’s National Grid could struggle to satisfy the voracious energy needs of AI and the systems it supports. The 62-year-old, who is Director of Data Cabling and Networks at Member firm OCS M&E Services, joins a growing number of experts who have warned about the new technology’s huge appetite for electricity, which is often greater than many small countries use in a year. And he questioned whether the UK’s current electrical infrastructure was fit for purpose in the face of the massive increase in predicted demand, not only from the power-hungry data centres supporting AI, but also from the continued rise in electric vehicle (EV) charging units. Mike says, “AI is becoming more embedded in our everyday lives, from digital assistants and chatbots helping us on websites to navigation apps and autocorrect on our mobile phones. And it is going to become even more prevalent in the near future. “Data centres, which have many servers as their main components, need electrical power to survive. It is therefore only natural that any talk about building a data centre should begin with figuring out the electrical needs and how to satisfy those power requirements. “At present, the UK’s National Grid appears to be holding its own, with current increases being met with renewable energy systems. But as technology advances and systems such as AI are introduced, there will be a time when the grid will struggle to support the demand.” Mike said that it is estimated that there could be 1.5 million AI servers by 2027. Running at full capacity, these would consume between 85 and 134 terawatt hours per year – roughly equivalent to the current energy demands of countries like the Netherlands and Sweden. He adds, “I remember attending an EV training session about 25 years ago and the standing joke was, ‘Where’s all this electricity going to come from?’ We all felt the network needed upgrading then, and now there is extra pressure from the new AI data centres springing up.” Mike has spent 44 years in the electrical industry, with 40 of those providing continued service at the same company; starting at Arthur McKay as a qualified electrician in June 1984, through to his current role at what is now OCS. He was confirmed as new SELECT President at the association’s AGM at the Doubletree Edinburgh North Queensferry on Thursday 6 June, taking over from Alistair Grant.

Scality RING solution deployed at SeqOIA medical lab

Scality, a global provider of cyber-resilient storage for the AI era, today announced a large-scale deployment of its RING distributed file and object storage solution to optimise and accelerate the data lifecycle for high-throughput genomics sequencing laboratory, SeqOIA Médecine Génomique. This is the most recent in a series of deployments where RING is leveraged as a foundational analytics and AI data lake repository for organisations in healthcare, financial services and travel services across the globe. Selected as part of the France Médecine Génomique 2025 (French Genomic Medicine Plan), SeqOIA is one of two national laboratories integrating whole genome sequencing into the French healthcare system to benefit patients with rare diseases and cancer. SeqOIA adopted Scality RING to aggregate petabyte-scale genetics data used to better characterise pathologies, as well as guide genetic counselling and patient treatment. RING grants SeqOIA biologists efficient access from thousands of compute nodes to nearly 10 petabytes of data throughout its lifecycle, spanning from lab data to processed data, at accelerated speeds and a cost three to five times lower than that of all-flash file storage. “RING is the repository for 90% of our genomics data pipeline, and we see a need for continued growth on it for years to come,” says Alban Lermine, IS and Bioinformatics Director of SeqOIA. “In collaboration with Scality, we have solved our analytics processing needs through a two-tier storage solution, with all-flash access of temporary hot data sets and long-term persistent storage in RING. We trust RING to protect the petabytes of mission-critical data that enable us to carry out our mission of improving care for patients suffering from cancer and other diseases.” Scality RING powers AI data lakes for other data-intensive industries. One of the largest publicly held personal line insurance providers in the US chose RING as the preferred AI-data lake repository for insurance analytics claim processing. The provider chose RING to replace its HDFS (Hadoop File System), and the customer has since realised three times improved space efficiency and cost savings, with higher-availability through a multi-site RING deployment to support site failover. Meanwhile, a multinational IT services company whose technology fuels the global travel and tourism industry is using Scality RING to power its core data lake. RING supports one petabyte of new log data ingested each day to maintain a 14-day rotating data lake. This requires RING to purge (delete) the oldest petabyte each day, while simultaneously supporting tens of gigabytes per second (GB/s) read access for analysis from a cluster of Splunk indexers. For data lake deployments, these organisations require trusted and proven solutions with a long-term track record of delivering performance and data protection at petabyte-scale. For AI workload processing, they pair RING repositories in an intelligent tiered manner with all-flash file systems, as well as leading AI tools and analytics applications, including Weka.io, HPE Pachyderm, Cribl, Cloudera, Splunk, Elastic, Dremio, Starburst and more. With strategic partners like HPE and HPE GreenLake, Scality has the ability to deliver managed AI data lakes. For more from Scality, click here.

Cirata to offer native integration with Databricks Unity Catalog

Cirata, the company that automates Hadoop data transfer and integration to modern cloud analytics and AI platforms, has announced the release of Cirata Data Migrator 2.5 which now includes native integration with the Databricks Unity Catalog. Expanding the Cirata and Databricks partnership, the new integration centralises data governance and access control capabilities to enable faster data operations and accelerated time-to-business-value for enterprises. Databricks Unity Catalog delivers a unified governance layer for data and AI within the Databricks Data Intelligence Platform. Using Unity Catalog enables organisations to seamlessly govern their structured and unstructured data, machine learning modules, notebooks, dashboards and files on any cloud or platform. By integrating with Databricks Unity Catalog, Cirata Data Migrator unlocks the ability to execute analytics jobs as soon as possible or to modernise data in the cloud. With the ability to support Databricks Unity Catalog’s functionality for stronger data operations, access control, accessibility and search, Cirata Data Migrator automates large-scale transfer of data and metadata from existing data lakes to cloud storage and database targets, even while changes are being made by the application at the source. Using Cirata Data Migrator 2.5, users can now select the Databricks agent and define the use of Unity Catalog with Databricks SQL Warehouse. This helps data science and engineering teams maximise the value of their entire data estate while benefiting from their choice of metadata technology in Databricks. “As a long-standing partner, Cirata has helped many customers in their legacy Hadoop to Databricks migrations,” said Siva Abbaraju, Go-to-Market Leader, Migrations, Databricks. “Now, the seamless integration of Cirata Data Migrator with Unity Catalog enables enterprises to capitalize on our Data and AI capabilities to drive productivity and accelerate their business value.” “Cirata is excited by the customer benefits that come from native integration with the Databricks Unity Catalog,” says Paul Scott-Murphy, Chief Technology Officer, Cirata. “By unlocking a critical benefit for our customers, we are furthering the adoption of data analytics, AI and ML and empowering data teams to drive more meaningful data insights and outcomes.” This expanded Cirata-Databricks partnership builds on previous product integrations between the two companies. In 2021, the companies partnered to automate metadata and data migration capabilities to Databricks and Delta Lake on Databricks, respectively. With data available for immediate use, the integration eliminated the need to construct and maintain data pipelines to transform, filter and adjust data, along with the significant up-front planning and staging. Cirata Data Migrator is a fully automated solution that automates Hadoop data transfer and integration and moves on-premises HDFS data, Hive metadata, local filesystem, or cloud data sources to any cloud or on-premises environment, even while those datasets are under active change. Cirata Data Migrator requires zero changes to applications or business operations and moves data of any scale without production system downtime, business disruption, and with zero risk of data loss. Cirata Data Migrator 2.5 is available now with native integration with the Databricks Unity Catalog.

Arista unveils Etherlink AI networking platforms

Arista Networks, a provider of cloud and AI networking solutions, has announced the Arista Etherlink AI platforms, designed to deliver optimal network performance for the most demanding AI workloads, including training and inferencing. Powered by new AI-optimised Arista EOS features, the Arista Etherlink AI portfolio supports AI cluster sizes ranging from thousands to 100,000s of XPUs with highly efficient one and two-tier network topologies that deliver superior application performance compared to more complex multi-tier networks while offering advanced monitoring capabilities including flow-level visibility. “The network is core to successful job completion outcomes in AI clusters,” says Alan Weckel, Founder and Technology Analyst for 650 Group. “The Arista Etherlink AI platforms offer customers the ability to have a single 800G end-to-end technology platform across front-end, training, inference and storage networks. Customers benefit from leveraging the same well-proven Ethernet tooling, security, and expertise they have relied on for decades while easily scaling up for any AI application.” Arista’s Etherlink AI platforms The 7060X6 AI Leaf switch family employs Broadcom Tomahawk 5 silicon, with a capacity of 51.2Tbps and support for 64 800G or 128 400G Ethernet ports. The 7800R4 AI Spine is the fourth generation of Arista’s flagship 7800 modular systems. It implements the latest Broadcom Jericho3-AI processors with an AI-optimised packet pipeline and offers non-blocking throughput with the proven virtual output queuing architecture. The 7800R4-AI supports up to 460Tbps in a single chassis, which corresponds to 576 800G or 1152 400G Ethernet ports. The 7700R4 AI Distributed Etherlink Switch (DES) supports the largest AI clusters, offering customers massively parallel distributed scheduling and congestion-free traffic spraying based on the Jericho3-AI architecture. The 7700 represents the first in a new series of ultra-scalable, intelligent distributed systems that can deliver the highest consistent throughput for very large AI clusters. A single-tier network topology with Etherlink platforms can support over 10,000 XPUs. With a two-tier network, Etherlink can support more than 100,000 XPUs. Minimising the number of network tiers is essential for optimising AI application performance, reducing the number of optical transceivers, lowering cost and improving reliability. All Etherlink switches support the emerging Ultra Ethernet Consortium (UEC) standards, which are expected to provide additional performance benefits when UEC NICs become available in the near future. “Broadcom is a firm believer in the versatility, performance, and robustness of Ethernet, which makes it the technology of choice for AI workloads,” says Ram Velaga, Senior Vice President and General Manager, Core Switching Group, Broadcom. “By leveraging industry-leading Ethernet chips such as Tomahawk 5 and Jericho3-AI, Arista provides the ideal accelerator-agnostic solution for AI clusters of any shape or size, outperforming proprietary technologies and providing flexible options for fixed, modular, and distributed switching platforms.” Arista EOS Smart AI Suite The rich features of Arista EOS and CloudVision complement these new networking-for-AI platforms. The innovative software suite for AI-for-networking, security, segmentation, visibility, and telemetry features brings AI-grade robustness and protection to high-value AI clusters and workloads. For example, Arista EOS’s Smart AI suite of innovative enhancements now integrates with SmartNIC providers to deliver advanced RDMA-aware load balancing and QoS. Arista AI Analyzer powered by Arista AVA automates configuration and improves visibility and intelligent performance analysis of AI workloads. “Arista’s competitive advantage consistently comes down to our rich operating system and broad product portfolio to address AI networks of all sizes,” says Hugh Holbrook, Chief Development Officer, Arista Networks. “Innovative AI-optimised EOS features enable faster deployment, reduce configuration issues and deliver flow-level performance analysis, and improve AI job completion times for any size AI cluster.” The 7060X6 is available now. The 7800R4-AI and 7700R4 DES are in customer testing and will be available 2H 2024. For more from Arista Networks, click here.

Arista announces technology demonstration of AI data centres

Arista Networks has announced a technology demonstration of AI data centres in order to align compute and network domains as a single managed AI entity, in collaboration with NVIDIA. In order to build optimal generative AI networks with lower job completion times, customers can configure, manage, and monitor AI clusters uniformly across key building blocks including networks, NICs, and servers. This demonstrates the first step in achieving a multi-vendor, interoperable ecosystem that enables control and coordination between AI networking and AI compute infrastructure. As the size of AI clusters and large language models (LLMs) grows, the complexity and sheer volume of disparate parts of the puzzle grow apace. GPUs, NICs, switches, optics, and cables must all work together to form a holistic network. Customers need uniform controls between their AI servers hosting NICs and GPUs, and the AI network switches at different tiers. All these elements are reliant upon each other for proper AI job completion but operate independently. This could lead to misconfiguration or misalignment between aspects of the overall ecosystem, such as between NICs and the switch network, which can dramatically impact job completion time, since network issues can be very difficult to diagnose. Large AI clusters also require coordinated congestion management to avoid packet drops and under-utilisation of GPUs, as well as coordinated management and monitoring to optimise compute and network resources in tandem. At the heart of this solution is an Arista EOS-based agent enabling the network and the host to communicate with each other and coordinate configurations to optimise AI clusters. Using a remote AI agent, EOS running on Arista switches can be extended to directly-attached NICs and servers to allow a single point of control and visibility across an AI data centre as a holistic solution. This remote AI agent, hosted directly on an NVIDIA BlueField-3 SuperNIC or running on the server and collecting telemetry from the SuperNIC, allows EOS, on the network switch, to configure, monitor, and debug network problems on the server, ensuring end-to-end network configuration and QoS consistency. AI clusters can now be managed and optimised as a single homogenous solution. John McCool, Chief Platform Officer for Arista Networks, comments, “Arista aims to improve efficiency of communication between the discovered network and GPU topology to improve job completion times through coordinated orchestration, configuration, validation, and monitoring of NVIDIA accelerated compute, NVIDIA SuperNICs, and Arista network infrastructure.” This new technology demonstration highlights how an Arista EOS-based remote AI agent allows the combined, interdependent AI cluster to be managed as a single solution. EOS running in the network can now be extended to servers or SuperNICs via remote AI agents to enable instantaneous tracking and reporting of performance degradation or failures between hosts and networks, so they can be rapidly isolated and the impact minimised. Since EOS-based network switches are constantly aware of accurate network topology, extending EOS down to SuperNICs and servers with the remote AI agent further enables coordinated optimisation of end-to-end QoS between all elements in the AI data centre to reduce job completion time. Zeus Kerravala, Principal Analyst at ZK Research, adds, “Best-of-breed Arista networking platforms with NVIDIA’s compute platforms and SuperNICs enable coordinated AI data centres. The new ability to extend Arista’s EOS operating system with remote AI agents on hosts promises to solve a critical customer problem of AI clusters at scale, by delivering a single point of control and visibility to manage AI availability and performance as a holistic solution.” Arista will demonstrate the AI agent technology at the Arista IPO tenth anniversary celebration in NYSE on 5 June, with customer trials expected in the second half of 2024. For more from Arista, click here.