CloudNets

Dudy Cohen, VP of Product Marketing at DriveNets, examines the rapid evolution of Ethernet technology and its critical role in AI infrastructure. Cohen discusses how Ethernet is advancing through multiple dimensions—from 800G to 1.6T line rates, co-packaged optics integration, and the introduction of scheduling layers that address Ethernet's traditional limitations. He explores the dramatic scaling of GPU clusters, from thousands to potentially millions of units, and explains why different scheduling approaches—fabric scheduling versus endpoint scheduling—suit different deployment scenarios. Cohen traces DriveNets' journey from building scalable service provider routers to applying the same scheduled fabric technology to AI backend networks. The discussion reveals how technical architecture choices directly impact business metrics, from time to first token to cost per million tokens, and why open, Ethernet-based solutions are gaining traction across hyperscalers, neoclouds, and enterprise deployments.What You Will Learn:How Ethernet technology is evolving across speed, optics, and scheduling capabilitiesThe difference between fabric scheduling and endpoint scheduling for AI networksWhy ESUN technology enables Ethernet to compete with purpose-built scale-up protocolsHow GPU cluster sizes are scaling from thousands to potentially millions of unitsThe business case for Ethernet-based AI infrastructure in terms of deployment speed and operational costs

From the SuperCompute25 event in St. Louis, this episode takes a look behind the scenes of the world’s largest temporary network. Hear about the real challenges of orchestrating massive, multi-vendor networks—and how a new platform aims to simplify network operations end to end, with big news coming in 2026.𝗟𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗖𝗹𝗼𝘂𝗱👉 https://drivenets.com/solutions/ai-networking/𝗪𝗵𝗲𝗿𝗲 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀🔗 LinkedIn:    https://www.linkedin.com/company/drivenets 🔗 X:   https://x.com/drivenets𝗣𝗼𝗱𝗰𝗮𝘀𝘁𝘀🔈 Apple: https://podcasts.apple.com/il/podcast/cloudnets/id1589786423 🔈 Spotify:  https://open.spotify.com/show/1q8e4hNOyaz5ySTeRjTP25

In this episode of CloudNets, we dive into the last-mile redundancy dilemma for service providers. We explore the trade-offs between single and dual connections, the complexity of traditional chassis-based designs, and how Distributed Disaggregated Chassis (DDC) delivers the best of both worlds—true redundancy, resilience, and scalability without added operational complexity.DriveNets Service Provider Networking solution: https://drivenets.com/products/Follow us on socialLinkedIn: https://www.linkedin.com/company/drivenetsX: https://twitter.com/drivenets𝗣𝗼𝗱𝗰𝗮𝘀𝘁𝘀🔈 Apple:  https://podcasts.apple.com/il/podcast/cloudnets/id1589786423🔈 Spotify:   https://open.spotify.com/show/1q8e4hNOyaz5ySTeRjTP25

In this episode of CloudNets, we dive into blast radius and how to design networks that prevent failures from cascading and impacting customers. We explore how network segmentation and distributed chassis architectures dramatically reduce blast radius and improve resiliency.𝗟𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗖𝗹𝗼𝘂𝗱https://drivenets.com/products/𝗪𝗵𝗲𝗿𝗲 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀LinkedIn: https://www.linkedin.com/company/drivenets X: https://x.com/drivenets𝗣𝗼𝗱𝗰𝗮𝘀𝘁𝘀Apple:  https://podcasts.apple.com/il/podcast/cloudnets/id1589786423Spotify:  https://open.spotify.com/show/1q8e4hNOyaz5ySTeRjTP25𝗪𝗮𝘁𝗰𝗵 𝗠𝗼𝗿𝗲 𝗩𝗶𝗱𝗲𝗼𝘀      • CloudNets S4 Ep7 Network Modernization        • CloudNets 4 Ep 1 New Automation Capabilities        • CloudNets 4 Ep 3 Autonomous Network  

This episode of CloudNets explains how network operators should prioritize upgrades by balancing new feature value against security, reliability, and SLA risks. It highlights the importance of addressing security vulnerabilities quickly using targeted hot patches, which reduce testing scope and operational risk while allowing upgrades to be scheduled and deployed with minimal network disruption.𝗟𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗖𝗹𝗼𝘂𝗱https://drivenets.com/products/𝗪𝗵𝗲𝗿𝗲 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀LinkedIn: https://linkedin.com/drivenets  X: https://x.com/drivenets𝗣𝗼𝗱𝗰𝗮𝘀𝘁𝘀Apple:  https://podcasts.apple.com/il/podcast/cloudnets/id1589786423Spotify:  https://open.spotify.com/show/1q8e4hNOyaz5ySTeRjTP25𝗪𝗮𝘁𝗰𝗵 𝗠𝗼𝗿𝗲 𝗩𝗶𝗱𝗲𝗼𝘀   • https://youtu.be/-8EbkT524yo?si=qs70VVUFBqrTQtki   • https://youtu.be/F7MwHpnbGP0?si=L5R-s3Kq4PgcyVOh   • https://youtu.be/SI5MPwFy_XM?si=MJAgiWkQl91vIH6Y

In this episode, we explain the challenge network operators face when introducing new features—balancing speed, risk, and maintenance windows—and why traditional “big bang” upgrades are risky and slow. Here, we introduce automated scheduled maintenance that allows gradual, controlled software rollouts, reducing risk while accelerating innovation without late-night manual work.𝗟𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗖𝗹𝗼𝘂𝗱https://drivenets.com/products/𝗪𝗵𝗲𝗿𝗲 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀LinkedIn: https://linkedin.com/drivenets  X: https://x.com/drivenets𝗣𝗼𝗱𝗰𝗮𝘀𝘁𝘀Apple:  https://podcasts.apple.com/il/podcast/cloudnets/id1589786423Spotify:  https://open.spotify.com/show/1q8e4hNOyaz5ySTeRjTP25

At the SC25 event, DriveNets’ Dudy Cohen (VP Product Marketing) and Andy Holland (Sales Director for AI Infrastructure) explore how Ethernet is expanding its reach beyond traditional scale-out and scale-across applications to challenge InfiniBand in the demanding scale-up domain.Key highlightsDriveNets powers the world’s largest AI and Telco networksDriveNets brings an AI Networking solution with peak performance out-of-the-boxDriveNets fabric scheduled Ethernet is solving industry-wide pain points in AI NetworkingWatch the interview Want to learn how DriveNets is reshaping AI networking infrastructure?Explore DriveNets AI Networking Solution𝗪𝗵𝗲𝗿𝗲 can you 𝗳𝗶𝗻𝗱 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀LinkedInX

What is the new service provider network? This is the network modernization process all service providers need to go through. Service Providers cannot continue to build networks the same way they have for over 20 years, and still maintain a valid business.

DCI is going through something and this something is called AI. The three changes that AI made to DCI are capacity, performance, and one environment and layer three capabilities.

What’s the importance of latency in AI networks. AI networks introduce new challenges that need different treatments of latency.

DES, DSF, DDC – Comparing Scheduled FabricArista recently launched their DES solution, the Distributed Etherlink Switch, which is essentially an end to end VOQ system – a large scale chassis. The approach is basically saying we should put as much logic as we can on the switch side. The switch will handle all the congestion control, all the reordering of packets, the load balancing of packets, which are all necessary for AI networking. So why choose DriveNets DDC over those solutions?

Are autonomous networks coming soon?We’re taking baby steps, starting with islands of automation and we’re going to automate small, mundane tasks by helping operators develop CI/CD pipelines that allow them to take innovation from their lab to a network digital twin and eventually into production in a much faster way.

We have an issue with resolving the AI fabric, or an AI networking problem with large clusters of GPUs usually used for training. This episode looks at the issue with resolving the AI fabric and explores how an Ethernet based solution can resolve this – by building a chassis which is distributed (a disaggregated, distributed chassis). This approach has no packet loss, is lossless and is a fully scheduled fabric, but without the scale limitation of a chassis.

What are the new automation sets coming to DriveNets Network Orchestrator?Autoboot profiles helps you bootstrap new white boxes from the factory. Smart Rollout is a brand new feature and it allows you to push software into the entire network after configuring just a few parameters. The Single Source of Truth integration seamlessly integrates the inventory that we have in DNOR into NetBox, or or other inventory systems.

Failure recovery is a very big issue when it comes to AI clusters because there are always failures and when the failure come, it’s a big thing because you need to stop the calculation, go back to the last checkpoint. You lose a lot of time and money and resources that are spent idle and and wasted time. And the networking part is crucial in order to create a fail.

Congestion demands attention, otherwise it can result in higher latency and packet loss. The main dilemma around congestion is around avoidance or mitigation. We’ll look at how scheduled fabrics make sure that AI infrastructure is lossless and predictable rather than bringing additional technologies to mitigate the congestion, but do not avoid it

Why is cell-based fabric is so much better than the alternative when it comes to Ethernet for AI fabric or AI infrastructure.

Chassis and Clos, which is better?From the perspective of scale, absolutely Clos topology outranks a chassis. From an operational standpoint, a Clos topology would mean that you will need to manage a lot of boxes and that's a bigger headache than managing a single device, a chassis, one stop shop. But from a Capex standpoint, a Clos topology prevails over chassis. Looking at the alternative, Distributed Disaggregated Chassis (DDC), it takes the same scalability factor of a Clos, the operational aspects of a chassis and even when you look at the cost side of things, from an operational standpoint, DDC operates as a single device. Then from a Capex standpoint, DDC operates or works as white boxes.

We're going to talk about KDDI, one of Japan's major Tier 1 operators. They recently announced that they implemented DriveNets Network Cloud solution in its network and they are very satisfied. They are talking about huge TCO advantages, about 46% saving in power and 40% savings in rackspace, etc. Let's hear about KDDI and why is this deployment, a major deployment, of Network Cloud so important?

How can a disaggregated network solution provide benefits in terms of Total Cost of Ownership to service providers in a long term perspective

What are three things we can learn about Network Cloud in real world network operations.

Today we're going to talk about recovering from major faults, those things that keep you up at night.

DriveNets hosted a panel for analysts and media at MWC23 in order to explain its work and hear from industry leaders on progress for disaggregated networking. Members of the panel included Igal Elbaz, SVP and network CTO at AT&T;  Jean Louis Le Roux, executive vice president for International Networks with Orange; Cayetano Carbajo, VP for core, transport and service platforms with Telefonica.

Today we're going to talk again about AI networking, and we will provide the solutions for the challenges we mentioned last time. With the fast growth of AI workloads, network solutions need to be ready to resolve issues including  having a flexible and online architecture, being able to  scale and maintain performance at scale, and  having a field proven rock-solid solution.

What are the challenges of AI Networking? We’re going to talk about AI, and specifically about AI networking. we want to talk about the challenges behind the AI infrastructure, because AI ML training, it’s very compute, intense task. What are the main challenges in AI networking that companies like the Hyperscalers that are going into this market now needs to resolve? 

We’re talking about some myths about disaggregated distributed chassis or Network Cloud. The myth about Network Cloud and this architecture in general is just a one off. The myth about the operational headache. And the myth about that it’s a fairly complex architecture.

We're talking about operations because usually when you talk to operations people about disaggregated and cloud-native systems, they do not like it. Since it's multiple vendors, it's multiple building blocks, it's a new thing to learn. And operations people will love to stick to what we have. But surprisingly enough, bringing the Network Cloud into your network actually means better operations, simpler operations.Watch CloudNets: https://drivenets.com/resources/cloud-nets/special-why-network-cloud-means-better-operations

DriveNets Network Cloud, changes the operational and economic model of the network, allowing it to scale capacity and services much faster while increasing service provider profitability.Watch CloudNets at https://drivenets.com/resources/cloud-nets/special-why-include-drivenets-in-your-next-network-rfp/

Now that we've established that building a Network Cloud cluster  is easy, let's talk about what happens afterwards because  the operations guys are the ones that having been doing the hard work all those years, after the engineering plans and installation.

How do you build a cluster with Network Cloud? The cluster is granular and built upon multiple boxes, so you can bring them with you, and find vacant places in existing racks to put it into and not have to have a dedicated rack for a new gigantic crowd.

Today we’re going to talk about the cloud, or specifically why cloud does not matter at all. You don’t need to worry or think about the underlying infrastructure, the hardware, the upgrades, the versions, the connectivity, etc, etc. And these are all transparent to you. This all looks like a cloud, something that you know that is there, but you don’t really care what happens in it. So this is why the cloud does not matter.

Today we are going to talk about the things we fear the most...End of life...which we, as network engineers, fear the most. So what do we do about end of life? With DriveNets' Network Cloud you have multiple hardware options, the entire cluster is built upon standard interfaces so it's very easy to replace or to upgrade the cluster with any box that adheres to those interfaces, and you can repurpose boxes in an exhausted cluster without the need for forklift. Watch the episode https://drivenets.com/resources/cloud-nets/season-2-ep-7-end-of-life

So what do we need to know about this move from 400 to 800?  800 is real. It's here, you can almost buy it.   It's a fairly smooth transition, having forward compatibility with the stuff you buy today.  The Network Cloud, the architecture, is agnostic to the interfaces or form factors. 

Today we are going to talk about DNOS, the DriveNets Network Operating System. Why is DNOS important and how it's different than just your ordinary network operating?

Let's talk about  ISSU, In Service Software Updates. What's new and how actually DriveNets can eventually make you upgrade your network without service.

We're going to talk about TIP DDBR, Telecom Infra Project Distributed Disaggregated Backbone Router and the RFI that went out. And hey, we got a ribbon, so we need to talk about it. Watch the episode.

Let's talk about optical modulation and some basic features of optics that us networking guys need to know.Watch the CloudNets episode

Let's about the optics that go into your router.  We're going to talk about ZR and ZR+ and open XR and stuff like that. So this is what you need to know about optics coming through the router: longer reach, leaner infrastructure, and lockout or the non lock in nonproprietary open system that allows to choose different vendors for your infrastructure.  Watch CloudNets https://drivenets.com/resources/cloud-nets/season-2-ep-1-optics/

Why is it important?As a newcomer into the networking space, the DriveNets disaggregated approach faced doubts. Explore how service providers have experienced the benefits of network disaggregation.Watch the video!

With regulators more and more talking about zero carbon emissions, going green and sustainability are high priorities. Service providers are feeling the push and need to put things into place in their operations. Why is sustainability so important for network operators?Watch the video!

Networks today need to scale, requiring more and more capacity. They also need to introduce new services in order to stay competitive and to satisfy the needs of the hungry subscribers. How do we deal with a data tornado?Watch the video!

Standardization is especially important for disaggregated networks, since software and hardware come from different vendors, the interconnect between those needs to standardized. But how does it reflect in the network cloud world? Watch the video!

Edge compute, edge networking, Telco Cloud. All those buzzwords mean one thing: the content, the essence of the network, the cloud instances are moving to the edge, to the edge of the network, to the edge of the cloud. That means they want to be closer to the subscribers.Watch the video!

Disaggregation of hardware and software, software based cloud-native instances, etc, these all have a dramatic effect on your total cost of ownership. So, when you go to disaggregation,  the Network Cloud, you save a lot of money. Watch the video!

Today, we're going to talk about high availability in the Network Cloud, because in spite of what you may think, the Network Cloud Infrastructure presents higher availability than the traditional chassis.Watch the video!

So third parties, multi-vendor: what are we talking about? Because DriveNets Network Cloud infrastructure can run not only the DriveNets applications or network functions, but also third party applications. Watch the video!

What is networking-optimized?It is taking networking functions and running them on networking-optimized infrastructure.Watch the video!

What is orchestration?Orchestration is the secret sauce that puts everything together in the distributed disaggregated system.Watch the video!

What’s multiservice?This is about running multiple network functions over the same shared virtualized network infrastructure.Watch the video!

Distributed disaggregation, why do we need it?Why is the disaggregation we know for years not enough?Watch the video!