This is the third and final post on the market segmentation relevant to Software Defined Networking and so we’ll be taking a look at the Enterprise Market. For our purposes we’ll set the bottom limit on Enterprises as companies with at least 50 employees, anything smaller would be classified as an Small Businesses. One of the most important difference between the Enterprise and the other two markets (Cloud-Scale Datacenter and Service Provider) is that the vast majority of enterprises don’t fit into any of the “philosophical camps” which define the schools of thought driving SDN. Fundamentally, most enterprises would like to see the network as a utility, in which price and availability were the primary inputs driving the purchasing decision. In reality networks are still far from a utility service, they require frequent reconfigurations (to support new modes of use) and upgrades (to increase capacity and support those new modes of use). For completeness, there are some large and/or specialized enterprises within which an organization may align with the views being held groups in the other two markets (in which the network is truly a strategic asset), but typically these most closely resemble the Service Provider datacenter players in their capabilities and their needs.
The reason that enterprises don’t typically align with the core SDN philosophies is that the atomic elements of SDN (bare metal infrastructure, OpenFlow, open-source controllers) are not intrinsically useful to the enterprise. A holistic solution (which may be built on those elements) which solves critical enterprise needs (specifically making the network approach the “utility” model of convenience and reliability and price) must be created to open the Enterprise market to SDN technologies. Of course, this market is served today by the subscribers to the “Establishment” school of thought – the traditional networking vendors. And a close look at the marketing messages from these vendors reveals that they have appropriated only half of the value proposition of SDN: the simplification of the user experience through automation and orchestration. However well intentioned, a sufficient simplification is unlikely to occur given the current proto-implementations, which are simply additional management layers laid down on top of the existing, complicated infrastructure. A sufficiently simple and elegant solution requires the abandonment of the multitude of redundant protocols and methodologies currently employed to implement existing network solutions. At the same time, this unlocks a simplification of the underlying infrastructure (access points, routers, switches) which could deliver the promised CAPEX reductions.
So why hasn’t SDN captured the Enterprise market yet? Those atomic elements (hardware, embedded software, controllers and end-user applications) simply aren’t available, or aren’t available with sufficient maturity meet the basic needs of enterprise (or service provider for that matter) customers. Without the maturity of those underlying components, the applications which can transform the operational experience of the end-users have no meaning. In the meantime, customers do get some incremental benefit from the focus the vendor community is putting on automation and orchestration, but it remains a very evolutionary, not revolutionary approach.
Ironically, network deployments are most susceptible to evolutionary approaches. One of the characteristics of cloud-scale datacenters that enabled the rapid adoption of SDN technologies was that they are built-out (and replaced) so rapidly. Essentially the datacenter is greenfield at the pace of server refresh (less than 3 years). Both enterprise and service provider networks are replaced at a much slower pace (typically closer to 10 years). In this sense the incremental approach that the vendors are taking is closely aligned with the existing customer base’s traditional procurement models. However, this leaves room for a truly disruptive solution which delivers both a significant CAPEX reduction as well as a transformational new operating model.
Leave a Reply