Monday, January 4, 2016

Blueprint: Four SDN Predictions for 2016

by Carolyn Raab, VP of Product Management at Corsa

In 2015, service providers, telcos and national research and engineering consortiums went through a major transition as they began implementing software-defined networks (SDN) to deliver programmable high performance and massive scale in the WAN and data center edge. And for network architects, operators and others involved in these next generation networks, the hard work is just beginning because the pressure will be on in 2016 to ensure that these SDN deployments live up to and exceed the hype. As these deployments move forward many architects will find themselves staring at a network that is completely different in size and shape than what they’re accustomed to. Fortunately, several new trends will help ensure greater control and scale across these networks, and compel us to make the following four predictions about the key developments that will benefit internet scale programmable networks in 2016.

1) FPGAs grow up and play a much larger role 

Network engineers need flexible, open hardware to create policy-driven, self-tuning networks. Hardware vendors need design cycles that can keep pace with network innovations and changes the network engineers demand.  FPGAs have advanced to the point where their underlying silicon process technology is in lock-step with ASICs, and can also provide users with the benefit of leveraging the combined volume of all other users of the same platform. They match the performance level and affordability of ASICs while offering full flexibility and rapid design cycles.  This shift to FPGAs will enable network architectures to evolve and scale more rapidly.

2) SDN will emerge from the hype cycle, based on real deployments

There are now confirmed, real deployments of SDN in service providers, Internet exchanges, ISPs, and data centers.  One challenge they all share is that the top-to-bottom solution requires an involved integration of SDN orchestration, control and data plane elements.  This clumsy stitching together of various parts of the equation has delayed real deployments as much as the lack of controllers and real SDN hardware that are performant and open.  However, with the availability of internet scale programmable and open hardware and open source controllers getting broad support the missing pieces are now present.  This top-to-bottom offering of interworking parts means real deployments will expand beyond the early, most sophisticated users to a broader base of networks of different shapes and sizes.

3) Re-programmable networks and real-time analytics will be hot topics for 2016

Because you program the network, you can make it better by creating an agile, self-tuning, automated network that create value for providers and users alike. This requires a virtuous circle of real-time statistics feeding into real-time analytics tools that trigger changes that are immediately programmed into the network.

To date, these tools existed, but in isolation of each other.  Now we see the beginnings of offerings that have created linkages to move towards closing the circle.  Through industry partnerships or as vertically integrated solutions from a single vendor, the ability to re-program the network on the fly is generating significant interest on the part of numerous stakeholders including service providers, broadcasters, municipalities, and enterprises.  All of them share a common requirement of needing to know what is going on in their networks so they can take the next appropriate action: Isolate? Allocate bandwidth? Add a new service?  Look for much discussion and some innovative deployments of re-programmable networks.

4) 2016: “The year of 100G SDN”

100G will begin to ramp up aggressively because both the data drivers and the underlying network have reached a critical junction.  Traffic growth on the network continues to put pressure on network infrastructure, and will be even more significant with 100G storage deploying to add to the massive growth in video and IoT generated traffic.  Operators will be able to answer with 100G SDN because of two key enablers:
  • Affordability – 100G SDN deployments are approaching a price point that is barely 3x what a 10G link would cost.  
  • Flexible feeds & speeds: QSFP28 for 100G, SFP+ for 10G and anything in between is possible with the same optics cage.  
Programmable SDN hardware designed with these cages can deploy as 10G initially and then rapidly move from 10G to 100G with a soft upgrade not a new hardware purchase to immediately address the data demands.

These and other trends highlight how large SDN deployments will require a more open and flexible approach at the software/firmware and hardware levels. It will be critical to ensure that networks can adapt and evolve as needed. We will be watching as networks take new innovative approaches to managing and orchestrating data in 2016

About the Author

Carolyn Raab is VP of Product Management at Corsa.

About Corsa Technology 

Corsa Technology is a networking hardware company focused on performance Software Defined Networking (SDN). Corsa develops programmable, flexible, internet-scale switches that respond in real-time to network orchestration, directing and managing traffic for SDN and NFV deployments from the 100G SDN WAN edge to networks needing full subscriber awareness. For more information, please visit


Post a Comment

See also