In the ever-evolving world of SDN Cisco, the network market-leader is…well…evolving. The latest story from the OpenStack event is that Cisco is in fact going to embrace OpenFlow and more mainstream industry principles, but my interpretation is that it’s also going to try to out-climb OpenFlow, to rise to the top of the stack to where the network and the cloud connect, and stake out some claims there while standards efforts are mired in the basement.
OpenFlow is to SDN what steering wheel is to car. You can’t have much of a car without a steering wheel, but by itself the wheel isn’t going to take you very far. SDN is about software defining networks, and there’s nothing in the standards yet that go anywhere near the software layer in a real sense. That software layer is the cloud, and more specifically the DevOps and virtual network interface the cloud implements. That’s why I’ve postulated three layers in SDN—the Cloudifier at the top to present services to the cloud, SDN Central in the middle to frame those services into network behavior, and the Topologizer at the bottom to interface bi-directionally with the network.
When VMware bought Nicira, they were taking a position at the cloud boundary and that’s what I think spurred Cisco into first picking up vCider and now tuning up its specific cloud support. They also seem to be doubling down on OpenStack, a smart move given that they have a server family that can run it and that OpenStack offers them a way to break with VMware (given that VMware is kind of breaking with them). So what Cisco seems to be doing is providing cloud-downward network control and virtualization that will incorporate OpenFlow where it must as a concession to standards but taking a lead where standards aren’t likely to move quickly. From there, Cisco can present their own approaches to lower-level control, and these may be richer than the standards can offer.
Cisco’s comments on SDN have come in the context of its announcement of an OpenStack distro of its own, and the adoption of this software internally in what some say is a replacement of VMware tools now in use. There’s a strong suggestion that Cisco is going to take the high road to SDN, literally, by controlling the cloud interface and the higher levels of SDN functionality, where application support is direct. This has the effect of shifting the focus from how SDN might impact current equipment to how SDN might impact the cloud, a much more favorable thing for a network vendor who’s trying to promote itself and its servers in the cloud space.
The significant truth here is that the cloud really is the driver of the network of the future, and the corollary is that you’re going to have a harder time as a network vendor if you don’t have an insightful cloud strategy. Cisco has the advantage of having servers; for other network vendors SDN positioning is doubly important because it defines their role in the cloud, a role that’s essential from a PR perspective. It’s not just PR, though. If you let cloud guys define the role of the network, they’ll define away all the differentiable features and you’re left pushing bits with Huawei looking over your shoulder.