Re/code is reporting today that IBM is pondering selling its Software Defined Networking (SDN) business unit. Unnamed sources are suggesting that $1B might be the price tag for the unit. This news comes only a week after IBM sold its x86 server unit to Lenovo for $2.3B. It's like the village fair at Big Blue with seemingly everything on the block. So what's up?
IBM is in the process of reinventing what it does - the purchase of Softlayer last year, and the increasing move towards Softlayer being the main thrust of its cloud computing offerings going forward indicate just how seriously IBM takes the battle with Amazon Web Services (AWS). While IBM is a little late to the party, it has the cash (or at least the asset backing), the enterprise credibility and the brand recognition to make up for a few false starts. It also realizes that no end of posturing or FUD will see it win this battle - it needs to offer credible products that customers actually want.
Recently, the company announced it was investing over $1B in building out a global data center footprint, aimed at competing directly with AWS service. It seems that losing the CIA cloud computing contract to AWS gave IBM just the rude shock it needed to overcome the inertia and start changing. That and Amazon CEO Jeff Bezos uttering the previously banned term "private cloud" sees IBM come directly into AWS' gun sights.
IBM's jeopardy-winning machine Watson is an exciting technology that, frankly, IBM hasn't known what to do with. Simply bandying around the term "big data analytics" doesn't really achieve anything. However, announcements in recent weeks about Watson technology being used across other IBM vertical offerings indicate that the company now understands it needs to provide a compelling business proposition for its customers - simply showing Watson winning Jeopardy won't do it. Enterprises need real solutions - analytics applied to everyday solutions to drive better results.
What we're seeing is the emergence of a new strategy from IBM that sees it only try and compete where it has a compelling differentiator. The server business is one which has been decimated by cheaper OEM vendors over the last few years - IBM is a top shelf vendor with pricing to match - it can't compete in the x86 server market so it was right to flick it on.
But is that the case when it comes to SDN? Is IBM right to focus on the higher level parts of the stack and leave the building blocks to others? It's interesting to reflect on the fact that VMware acquired SDN vendor Nicira a year or so ago for $1B, and now IBM tries to sell its own SDN asset for a similar price. The answer to this questions lies in one's perspective on what future enterprises will look like. If they will look much like they do today, with IT needing to assemble a number of different "plumbing" elements in order to build a series of services, then SDN is a critical part in that process. If, however, the enterprise of the future sees organizations taking predefined blocks of service, all of which have the raw infrastructure elements abstracted away from the end user, then flicking off SDN makes total sense - if you're selling a vertical specific, Watson-powered, big data crunching solution, who cares about the flexibility you have in your networking topology underneath?
Overall, this new, higher value strategy makes sense given the sort of organization IBM is. Its size means it's unlikely to win and price battle and lacks the agility to compete with more nimble commodity vendors. Whether there's enough value up the stack to support the monstrous size of IBM's organization, only time can tell.
