On 14 February 2013 we announced ExtremeXOS version 15.3. The latest version of our robust, modular, edge to core operating system features a number of enhancements including support for AVB (Audio Video Bridging), Identity Management enhancements, XNV Dynamic VLANs and GRE Tunneling enhancements, but some are saying that the really big news is support for OpenFlow and OpenStack Quantum, which are key to standards-based SDN support.
With widespread acceptance of the idea that there is value in having centralized, standardized command and control separate from the switches (and routers) that comprise the network comes the opportunity for developers to add value with SDN applications. The way this is going to unfold is going to be a lot like the transition we saw when mainframes, which were tightly closed, vertically integrated things where everything from soup to nuts came from the same vendor gave way to x86 machines. In this new world, no longer was there complete integration with everything from hardware to cables to software all coming from the same vendor, instead what happened is you had some like Intel and AMD building CPUs while others did motherboards. Folks like Microsoft, Apple and a variety of open and not so open *nix shops provided the OS while independent developers, who were far closer to their customers than anyone else, built the applications.
Similarly in the mobile world, before smartphones there were feature phones where whatever functionality or apps the phone was going to have shipped with it from the factory. This approach ultimately didn’t scale. With the rise of the smartphone and things like the App Store or Google Play we have some players building hardware, others doing the OS and another group building apps. This has taken us from a kind of bare shelves socialist purgatory to a vibrant and colorful world filled with Angry Birds, Pandora, Yelp and other games and apps.
In the old world of networking, prior to SDN, if a customer wanted some sort of functionality that did not exist in the switch, the customer would talk to the vendor and request a feature be added. The feature may or may not be added, but even if it was something the vendor wanted to do, it would usually take 12-18 months before that feature or capability would ship – if it ever shipped. The new world of networking, networking in the SDN era, will start to look a lot more like the world of Macs and PCs or the world of smartphones than the world of mainframes.
Some, like Broadcom, Marvell, and Intel, will make the merchant silicon that, like the x86 CPUs in Macs and PCs, power the network gear. Others, like Extreme Networks and some of these other network vendors, will build the network OSs, the platforms used to not only run the network, but just as importantly, the platforms used to run the apps. Just like there is an ecosystem of software developers for Windows and Mac and iOS and Android, there will be an ecosystem of software developers who build the SDN, OpenFlow and other apps that will be an important part of the networks of the future.
One thing that should be addressed is commoditization. Many in the industry have said that SDN will lead to commoditization of network hardware. While it might lead to the commoditization of the hardware produced by some vendors, the fact remains that underneath all the virtualization (it isn’t turtles all the way down, after all) networks are going to have to run on some sort of hardware and in many of these networks that hardware is going to have to be pretty special. Indeed, SDN is likely to create opportunities and allow network applications to potentially run and run well over multivendor networks. One might even suggest that rather than network commoditization, what we are going to get instead is network liberation and a blossoming of interoperability, ideas and network apps. Such is the nature of open standards and such is the nature of SDN and approaches like OpenFlow and OpenStack.
Speaking of network apps, there is an appstore for that. We have a community, xKit, where users can download core applications and automation tools. Still a humble beginning, xKit and similar resources point to a future where, like the pc and smartphone, software is a major differentiator.
We still have another 11 months before the end of the year, but if the trends we have seen so far continue, 2013 is going to be a good one.