Cisco’s restructuring to name two “presidents” (Gary Moore, president and COO, and Rob Lloyd, president of development and sales) is raising yet again the issue of succession, but the question of who runs Cisco is less significant to the market than where he runs it. Bloomberg said “While the discussion over succession indicates a new openness to change at Cisco, it may be instrumental to winning back business….” Is there any beef in that? Do they think Chambers is holding Cisco back? If so, will Moore or Lloyd do something different?
Cisco is critical to the direction of the market right now, for a number of reasons. First, they have the most successful account control strategy in networking so they drive the market more than any other vendor. Second, because of this, they are more at risk to the submergence of networking into the cloud, and the measures they take to try to stay on top will set the sales agenda of the industry. Third, their adherence to the “fast-follower” strategy for the last decade has limited innovation in networking, but arguably maintained IP as the de facto direction of the market. Will they abandon that?
So how does it look? To me, the vCider deal might have been the critical announcement. As I said last week, vCider isn’t an SDN player, and Cisco’s plans for it (as they state) involve binding it into their Quantum interface implementation, which means it’s the low end of the OpenStack cloud abstraction of the network. If Cisco was intending to chart a new course with the naming of two new presidents, then it would make sense to start sailing in that direction with any concurrent M&A activity and even announcements. The vCider deal seems to indicate that Cisco will work to tune the network to the cloud’s interfaces, which means they don’t plan to try to drive either SDN or the cloud in a new way.
Interestingly, vCider and Quantum pose almost the same risk to Cisco that full-on SDN might. The problem with SDN isn’t OpenFlow or standards, it’s the notion of software control of the network, which implies software VENDOR control of the fusion of networking and IT into the cloud. The big step, the missing link, is the notion of a flexible abstraction of network services that can be used by applications (the cloud) to define their service needs and used by hardware (the network) to drive device behavior in some way. Quantum seems to be moving in that direction, something I’ve also been noting here in my blog. The question is whether it’s moving fast enough and far enough, and whether Cisco is inadvertently advancing its cause by supporting the interface with vCider. The key contribution of SDN isn’t the OpenFlow management of forwarding tables, it’s the still-undefined abstraction that links virtual network SERVICES to actual device COOPERATION. Whether that cooperation is coerced by manipulating systems of devices through today’s service protocols and management systems or driven down explicitly through OpenFlow, we still end up with software driving the bus.
The thing is, Cisco has no real option. Virtual networking is out of the bag, and so Cisco can either embrace it and speed the devaluing of its incumbency, or resist (even obstruct) it and slow that reckoning for at least a time. We know what they’ll likely do, and competitors in the IT world will likely be happy to let them do it. To try to absorb the network into the cloud explicitly from above creates pressure on Cisco to climb into the cloud and compete strongly there, which isn’t what the vendors want. So the logjam here will have to be broken by a more pure and focused cloud or network player, somebody who can gain advantage not by being in both places but by being at the place where they combine. That’s what I hope will emerge from the fall’s SDN announcements, because SDN is that middle.
Juniper is the classic Cisco rival and has figured in speculation about a big network-vendor M&A, but I’m not sure that this is their time. UBS just released a note on Juniper that asks “Can it execute on its product transitions?” and that cites the critical role that PTX and QFabric will have to play. I still believe that Juniper booted both launches last year because they wouldn’t position either product correctly—meaning as the foundation of a cloud-distributed data center or the core of “cloudnet”. Juniper is still pushing boxes uphill, each in its own direction, as the market demands convergence of network concepts on (yes, you guessed it) the cloud!
What’s the value of a single-source provider for the cloud, like Juniper, versus the combination of players like F5 and Palo Alto cited by UBS, if Juniper isn’t going to integrate its own stuff around a strategic vision? They have TWO MONTHS to get this right, after which I don’t think any effective positioning can be gained. Then the torch passes to another competitor.