From a news perspective, a year is defined by the stories that make headlines. But markets are frequently moved more by subtext than overt story. Its sometimes the less sexy and more subtle happenings at the fringe of media interest that determine how industries will evolve and ultimately who will win and lose. This past year was certainly the year of software-defined networking, but what important supporting bits played out just out of the limelight?
The Launch. Given the amount of industry build-up, including rumors, a couple of head fakes, at least one aborted launch, and a competitive response that included rushing another launch, Insieme’s coming-out party will from here on after be known simply as The Launch. When the Big Dog invests north of $850M, heads will turn. But amidst all the product talk and industry dynamics, the bigger story seems to have been missed.
Arista pulled its own launch forward, ostensibly to take advantage of the media cycle that week. Whether they were trying to one-up Cisco or simply ride the wave is unimportant. The effect was to draw media attention down into the bowels of the networking industry. With two launches the same week, the press was almost forced to compare the companies, pulling what should have been a higher-level discussion back into the weeds – a little bit of marketing crabs in a bucket, I suppose.
But the bigger story was not how Cisco competes with Arista. All year, John Chambers has been talking about Cisco’s real endgame: being the number one IT company in the world. It’s not about networking; it’s about all of IT. For Cisco to take that position, it means they will need to take share not from Arista but from Oracle. Amidst all the noise about switches and cabling (!), a good chunk of the industry missed the fact that two heavyweights are strapping on gloves and warming up. Remember that this comes not too terribly long after Oracle made a couple of acquisitions (Xsigo and Acme Packet) of its own that start to take aim at more than just enterprise software.
Consider also that Cisco ended up launching right past SDN and into application-centric infrastructure. Forgetting for a moment that infrastructure is broader than networking, note that the marriage of applications and infrastructure will at some point involve the applications themselves. Cisco has long had eyes on building a software business. Ten years from now, we might be look back on 2013 as the year that the real fight for IT began.
Software-defined networking and hardware? With all the buzz around the S in SDN, it was easy to miss some of the more subtle happenings on the hardware side. The current generation of switches seem to be all but converged on Broadcom silicon. This has certainly been a big year for Broadcom’s Trident-II as it is featured in launch after launch.
But all of those design wins for Broadcom create an almost irresistible opportunity for Intel. With Fulcrum already in the stable, how long until Intel makes its push for capturing some of that switching silicon market?
At the Open Networking Summit in April, Stanford Professor and OpenFlow inventor Nick McKeown delivered an entire keynote presentation on the role of silicon in SDN. He outlined on stage what an SDN-optimized ASIC might look like. He might have been talking to a room full of engineers, but in that moment he was addressing Intel. If 2013 turns out to be the beginning of a rash of SDN deployments, it is not inconceivable that Intel might use this transition to take a serious run at the silicon market.
Broadcom certainly knows this, and they appear to have been setting up their own defense. Another big trend in networking is white box switching. And in this space, one of the more interesting efforts is Facebook’s Open Compute Project. Facebook consumes infrastructure unlike virtually any other company in the world. They turn gear over on much faster refresh cycles, which means CapEx is a huge issue. Accordingly, they are spearheading an initiative to produce a white box reference platform. If this is successful, you might see companies stamping out thousands of devices based on this design. At the center of the reference platform? Broadcom has positioned its switching silicon. Your move, Intel.
The Snowden Effect. When Edward Snowden exposed the reach of the NSA’s monitoring efforts, shockwaves rattled across the globe. Companies whose chief asset is data were forced to tap dance, backpedal, and in some cases even shut down. But beyond what we have already seen, there could be some very industry-specific impacts in networking.
If consumers do not believe that their data is safe in clouds that are located in the US or provided by US carriers, there might be a shift in buying patterns. It is possible that European and Asian cloud providers seize on this momentum to create offshore clouds to address data-sensitive services. This would pull infrastructure spend offshore, changing not only the buy side but also the sell side.
If SDN is designed in part to fuel the cloud, then a host of SDN startups could find that their buying base is no longer located entirely in North America. When to expand is always a difficult question to answer. This type of move could force the issue sooner than companies would like. Does a shift like this level the playing field for the more entrenched vendors? It’s impossible to know, but it adds a wrinkle to what had previously been largely a technology discussion.
Years from now, will we look on 2013 as having been defined by the stories or the subtext? It’s impossible to predict. But no one can argue that there isn’t a lot going on, both on stage and behind the scenes.
[Today's fun fact: Henry Ford produced the model T only in black because the black paint available at the time was the fastest to dry.]