Solving the “Undiscovered Country” Dilemma for Hyperscalers and Cloud Service Providers
(Tom Herbert, SiPanda CTO, April 22, 2021)
Most of us that conceptualize and develop products for the IT ecosystem are trying to
solve real problems for customers, and logically the only way to understand those
problems is to ask the customers; but, as Henry Ford (dubiously) said: “If I had asked
people what they wanted, they would have said faster horses.”, even when customers
are able to articulate their problems they often don’t suggest the best solution.
Hyperscalers and cloud service providers (CSPs) are no exception, and in fact the
challenge takes on an extra order of magnitude of difficulty and obscurity. Hyperscalers
and CSPs are (generally) notoriously secretive about their IT infrastructure, and for
understandable reasons – it is that “secret sauce” which is often their competitive
advantage. Not only that, they are motivated to continuously evolve their infrastructure
which means that tomorrow’s problems may be quite different from today’s, so even the
hyperscalers and CSPs themselves might not fully know the problems. It is kind of like
the movie “Star Trek: The Undiscovered Country” – you don’t know what you are up
against and you don’t know exactly where things are going tomorrow that will
necessitate significant replanning. So the question is: how do you provide solutions for
such an ecosystem with all these “undiscovered country” unknowns?
The simplest answer to this question
is to fallback to first principles. If you
want to build network datapath hardware and software for hyperscalers and CSPs, you first start with what you know your customers (hyperscalers and CSPs) generally need and where the industry is generally headed. We believe the requirements are a network core capable of over 1Tbps with wire-speed protocol processing performance and full software programmability. The emphasis on the word “full” is intentional as full easy-to-use software programmability is in its nature the best avenue for solving the unknown and future problems; we have known this in the computer industry for at least half a century and it’s clear that needs to be applied to all levels of networking infrastructure. However, full programmability isn’t enough, we also have increasingly demanding performance requirements with seemingly insatiable demands to move more and more data at lower and lower latencies. Current conventional wisdom dictates that flexibility and high performance cannot be simultaneously achieved; to get highest performance flexibility must be sacrificed and vice versa. At SiPanda we pay no homage to conventional wisdom-- we believe the flexibility-performance tussle is resolved with our Domain Specific Architecture that combines a very powerful, comprehensive, and ubiquitous programming model, PANDA, with a hardware platform, Panda-V, that is specifically tuned to accelerate and optimize the PANDA-programmed datapath. This unique combination of software and hardware enables the networking datapath platform for hyperscalers and CSPs to address their “undiscovered country” problems both today and tomorrow.
SiPanda was created to rethink the network datapath and bring both flexibility and wire-speed performance at scale to networking infrastructure. The SiPanda architecture enables data center infrastructure operators and application architects to build solutions for cloud service providers to edge compute (5G) that don’t require the compromises inherent in today’s network solutions. For more information, please visit www.sipanda.io. If you want to find out more about PANDA, you can email us at firstname.lastname@example.org.