SDN: A Study in Action and Reaction
Every action has consequences. In the world of physics, this usually entails an equal and opposite reaction. In human endeavors, however, the consequence is rarely equal or opposite, and quite often comes about so slowly that its true cause is subject to debate.
Deployment of software defined networking (SDN) is an action if ever there was one. That it will have a dramatic effect on data environments and enterprise productivity is beyond doubt. But the question is, how? At the moment, predictions abound about all the ways SDN will improve enterprise networking, but very few prognosticators dare to predict what challenges still lie ahead. And if past is prologue, can we expect software defined networking to solve a wide range of problems that enterprises encounter in their physical networks, only to replace those problems with even thornier issues on the abstract plane?
The unknown consequences of SDN in the data center
The simple fact is that we don’t know what the ramifications of SDN in actual production environments will be because there is so little field experience. According to Gartner, barely 4 percent of attendees at the recent IOM conference in Orlando, FL reported actual production deployments so far, and three quarters of them describe their activities as “limited.” More than 75 percent of the overall gathering reported little to no knowledge of SDN or are in the very preliminary stages of evaluation. In fact, Gartner reports that a high degree of skepticism follows software defined networking, with many networking executives still unconvinced that it can impact their data environments to a significant degree.
And while SDN may make the network more manageable, it does little for the applications that actually fuel data productivity. In fact, it may make app management even more difficult because the traffic path is now so highly automated, according to management software developer WildPackets. There may also be a plethora of networking devices from various vendors, and even though they conform to whatever SDN protocol is in place, troubleshooting pain points will still be a hassle.
SDN: Not the data center endgame
Enterprise executives also need to consider that software defined networking is not, in fact, the last piece of the virtual, dynamic data center puzzle, says Michael Bushong, VP of marketing at Plexxi. Optical transport, photonic switching and automation are all needed to confront a number of logistical problems and other issues that arise in fully virtualized environments. How do you maintain performance when harnessing resources separated over great distances? How do you manage a programmable fabric, particularly one that could be global in scale? How do you keep track of virtual resources, which may be hosted on one device and then another in the blink of an eye? SDN, it seems, is not the culmination of the virtual transformation, but merely a waypoint.
And all the while that the network is becoming more software defined, computing infrastructure is converging, notes SDN Central’s Craig Matsumoto. How, exactly, will SDN function when compute, storage and networking are converted to container-based modules? Companies like Docker and Google are already working on new architectures that enable applications to be divested across multiple containers and then employed for various workloads using highly scalable orchestration schemes and protocols. Not only is this breaking down the way we think of networks and infrastructure, but it is leading to entirely new concepts regarding the use and management of computing power and the ways abstraction can be used to leverage formerly rigid data parameters.
Of course, it’s always easier to view technology in terms of how it solves the problems of today rather than what it can do to remake the future. The enterprise has only just begun the SDN transition, so it is fair to say that no one truly knows what the results will be.
Photo courtesy of Shutterstock.