Wow. Just, wow. Here we were at Pica8, ten days away from announcing a screamingly simplified network switching architecture for enterprise campus networks – one that makes legacy switch stack and chassis switch replacement with disaggregated network switches ridiculously easy — when Arista suddenly pops up and says that the success network switches are having in the data center is now well and truly annoying to them, so they are starting to plan a future expedition to the enterprise campus in search of replacing lost revenue.
But after blowing straight past total market validation for us – we already have a number of ongoing enterprise campus network deployments inside some of Cisco’s oldest and largest accounts – Arista then went on to articulate the changes they envision for their future campus network switching push and, in doing so, came very close to a one-to-one mapping with what we’ve just announced as an orderable solution with our new PicaPilot switch orchestration software (see link below). (I really must order a case of champagne for my friends over there.)
Just how close is the mapping? Well, let’s start with solving the urgent requirement to simplify today’s legacy enterprise campus network architectures. (Sound familiar?) In our case, we took the web-scale data centers’ proven leaf-spine architecture, extracted all its complexity, and made it look like a single-layer flat network managed by a single IP address. Like I said, simple as can be. Add to that things like automatic rollbacks, increased network security – hey, we were the first in the business to ship a Linux NOS on a flexible, network switch – and software-driven network control, and it’s like we’ve stumbled into Mr. Peabody’s Way-Back Machine to Arista’s announcement earlier this month. The mapping isn’t perfect, of course, but nothing is. (Even Beethoven screwed up on even numbers.)
Clearly, there will be more than enough room for Arista in the enterprise switching market once they finally join us here, if for no other reason than they appear to be targeting Cisco’s huge enterprise core switches with a repurposed data center switch – makes sense – while we are burrowing into the market from the access-edge, which is where all the branch and campus move to disaggregated switching is actually taking place. Who knows? We may even end up meeting them somewhere in the middle. Might be nice, actually.
That said, even before our PicaPilot launch, we were already leading the bare metal hardware movement to replace enterprise campus networks. Not only are we the only networking vendor of any stripe to support both L2/L3 and OpenFlow (SDN), but we also have a complete enterprise feature set – support for voice VLANs, legacy Cisco phones, dumb phones, OSPF, PVST, NAC, QoS, PoE management, and a laundry list of capabilities that simply don’t exist in the data center — baked into our PICOS NOS. Without all of this, it’s simply not possible to stand up to Cisco in an account. Arista well understands this requirement as their announcement included a future partnership with Aruba — expensive, not flexible, and proprietary, but sporting an enterprise feature set – to provide “complete-solution” air cover for their future campus expedition.
The only real nit that I have about our co-joined announcements is that while we chose to go with simple nomenclature – selecting “pilot,” as in letting the enterprises pilot their own deployments — Arista elected to go all high-concepty and marketecture-ish with theirs, calling it the Cognitive Cloud Campus (with a side order of behavior-driven actions). How this conceptually gets to the heart of the main problem we’re both attacking – the need for simplicity – does seem a bit of a head-scratcher. Just sayin’.
So, Arista, welcome to our world. We’ll leave a light on for you.
Take a peek-a at our PicaPilot announcement here: Press Release