I was in the audience for a recent NFV focused industry event, and watched a product manager from a large incumbent vendor present an all-in-one solution for NFV Infrastructure (NFVI). He did a good job for his employer. However, the next presenter, an architect at a prominent tier-1 communication services provider, started his talk with “...and that single vendor stack is exactly what we are trying to get away from. NFV is, for us, a structural path to best-of-breed.”

Today, NFV Infrastructure at scale is hard. A number of disciplines ranging from OpenStack automation to hypervisor tuning, from Linux networking to data center leaf/spine design, get involved in architecture decisions. Having been a part of one of the industry’s largest and most successful NFV deployments, our team at Big Switch has seen this multi-disciplinary problem first hand. The stakes are high. An unexpected performance bottleneck or a single point of failure can wipe out the economic advantages that were the catalyst of moving to NFV in the first place.

In light of the engineering complexity of NFVI, some communication service provider teams retreat to the comfort of their incumbent vendor’s all-in-one solutions. Their executives abandoning engineering prowess in favor of procurement prowess. They don’t look for a best-of-breed approach.

But, there is a different path.  

At Big Switch, we are believers in Gartner’s, recently espoused position: Shift Network Spend From Premium Products to Premium People. We believe this is particularly true with carrier grade NFV deployments. Thanks to the close collaboration with our early NFVI end users, we identified 25 different optimizations for switching and routing in NFVI back in 2015 - our first year in NFVI trials. These came from CSP NFVI engineers that worked across multiple disciplines, connecting the dots between the virtual network functions, host networking, leaf/spine networking, automation systems and beyond.

We have long surmised that many of the extraordinarily talented, multi-disciplinary NFVI engineering teams with whom we work are outpacing incumbent vendor NFVI engineering teams. It was heartening to see the results of a recent ACG report, Creating Agility & Efficiency at Scale: The Economic Advantages of Open Architecture Platforms In NFV Deployments. This report compared two designs side-by-side in production. Caveat lector - Big Switch was actively involved in this research. Our involvement gave us access to the voluminous excel model that came from interview after interview with members of architecture, engineering and operations teams looking over these deployments.

ACG’s conclusion was that deployments of the best-of-breed NFVI pod design resulted in an estimated 53% reduction in total cost of ownership over the period of five years AND that delivered activations of new service offerings 3x faster than the incumbents’ alternative.  

For a CSP to embark on the best-of-breed path to NFVI, it is necessary that its engineering team has the multi-disciplinary chops to go after the technical challenge, and that executives support the team through the process. When strong engineers are set up for success in NFVI, the results speak for themselves. To me, the most important take-away of this side-by-side comparison is the ROI that comes from investing in great people.

It is an exciting time to be in networking.


Kyle Forster