Growth of Open Networking: Bringing Disaggregated Networking into the OCP

Blog-Run-Growth-of-Open-Networking

Ever since AT&T submitted their Distributed Disaggregated Chassis design to the Open Compute Project, there’s been fascination around how this development will influence the industry. Now that AT&T has acknowledged that DriveNets is indeed providing core-networking routing software for its next-gen core network, based on the DDC design, there’s a lot of excitement around DriveNets’ distributed disaggregated solution.

Despite being a relatively new community member, the OCP was excited to talk to DriveNets, seeing us as really leading the way in open networking. “OCP opens up a lot of opportunities for anybody dealing with disaggregation,” I  explained in my recent interview with Steve Helvie, VP of Channel Development at the Open Compute Project Foundation (OCP).

“Well, I think you guys are poised to capitalize on the growth of open networking,” said Helvie, “supposed to be close to 1.3-1.5 billion dollars by 2023 and we’re expecting DriveNets to lead the way in some of the software that’s being certified and validated and being used within open compute.” 

DriveNets Network Cloud enables a disconnect between the network growth and the network cost. This makes sense when it comes to replacing equipment to expand bandwidth, but adding new functionality shouldn’t mean replacing the network as well. This is part of the innovative capabilities that allows Network Cloud to enable running multiple services over the same infrastructure. 


To Learn More, Download the White Paper: Introducing Network Cloud
Transforming Service Provider Networks

Download

Applying the Circular Economy

As the amount of hardware in network data centers continues to grow, the OCP has shown a commitment to looking at ways to help reduce waste and embodied energy across all areas of the network infrastructure, with OCP members leading the effort to support a Circular Economy approach.

Run explained that the DriveNets approach enables a whole new aspect of the circular economy approach. “Outdated devices can be pushed out of the network and repurposed, either within the same network or even elsewhere, because with DriveNets Network Cloud the entire functionality is controlled by software and the only reason to add boxes is for expanding capacity.” 

This way existing hardware utilization is maximized, for greater impact not only on the network TCO but also on the environmental cost. 

DriveNets is doing a lot of work with AT&T

I explained to the OCP how we took a fully functional router and made adaptations to have it run on white boxes. “Eventually, what we ended up with is a solution, which is as flexible as software, while running on the server and as robust or capacity capable as what you would get running on a dedicated hardware silicon machine.”

In other words, this offers the best of both worlds and AT&T was our primary target. Of course, AT&T had their agenda of promoting disaggregation, disaggregated solutions and in this case, it was targeted for their core network. Once AT&T made the announcement that they are implementing disaggregation from DriveNets, it was already after we were in their production network and serving the core router with our clusters, which support about 200 terabits per second, the largest capacity that we installed into AT&T’s network. It goes without saying, you cannot go into AT&T without being carrier-grade, and without going through a rigorous testing process by the AT&T team. Now we are working with AT&T on expanding to further core locations, as well as to other network use cases.

A very big router

Core routers typically need to have many, many, ports supporting very high capacity – all running smoothly without dropping connections. “This is not a peering device, but in terms of capacity, in terms of bandwidth, it’s very challenging. The solution is implementing a core router based on an array of white boxes connected in a fat-tree topology,” I explained. For DriveNets, we have two types of boxes: a front white box which operates like a line card as in a chassis and a fabric line card, which operates like the backplane of a chassis. All line cards are connected to all fabrics, and all fabrics are connected to all line cards, just like in a chassis. The whole thing is orchestrated by our Network Cloud software. We have two pieces, our Network Operating System, known as DNOS and our orchestrator, known as DNOR. DNOS provides the actual functionality from the bottom up and DNOR provides everything to do with orchestration from the top down. 

“From AT&T’s perspective, it’s a very, very, big router. Of course, other use cases vary, and the size is not always the key item, sometimes it’s the service scaling that is the reason to go with this kind of a solution. But the pillars remain the same pillars, the building blocks are the same building blocks. We’re still using the same two types of white boxes, the software is the same DNOS and DNOR, the ecosystem that we’re working with is the same ecosystem; it’s just that the use cases can change.”

What’s next for DriveNets in open networking? 

DriveNets platform is not just focused on routing, but is laying the groundwork for multi-service capabilities. While routing is a very dignified service on its own, it’s not the only service which a network needs to address. Anything to do with DDOS prevention as a function, firewalls, DPI, load balancing, additional network functions which would have originally been hosted on an appliance, can now be hosted on these white boxes and orchestrated by the same software tools that you have within DriveNets Network Cloud. 

“We are in the process of launching these new services, not just those coming from DriveNets, all of the routing services that I mentioned before have been developed here, in-house. When it comes to other network services, there are expert companies dealing with this and we are establishing these relationships with the third-party providers to have their services containerized and run throughout the network.”

It makes a lot of sense to have a firewall located on the network device itself and not in the back office on a server. This offers a lot more flexibility and better control over what’s actually happening and what’s actually running within the network. That’s just an example of a well known service, but looking forward to the launch of 5G networks – data-driven services – there will definitely be a need for instantiation of these services within the network. This is exactly what DriveNets Network Cloud offers service providers.

About the author

Run is the Head of Product Strategy for DriveNets. Previously, he ran the ethernet switch product management and marketing teams at Mellanox Technologies, playing a central role in promoting disaggregation for data center networking. At Nokia Siemens Networks, Run served in the packet networks business unit. Run holds a BSc. in Electrical Engineering and Computer Science from the University of Tel Aviv, Israel.

INTRODUCING DRIVENETS ORCHESTRATOR:

DNOR Accelerating the transition to cloud-native networks

DriveNets White Paper
Bringing the Simplicity and Visibility of Cloud Orchestration to Network Cloud

DNOR-White-Paper
Skip to content