SDN Presents a Catch 22 For The Enterprise

Lowering the cost and complexity of enterprise network management - an alternative to SDN?
– Subhasish Gupta, Country Manager – India & SAARC, Allied Telesis

Software Defined Networking has moved into the spotlight recently. With claims of better network utilization, this centralized management promises the big advantage of treating the network as a single, virtualized device.
SDN detaches network control from the hardware and moves this into a software application, where the server defines network parameters. In essence, it’s ‘clever quality of service’. Prioritize the right data over the network and make sure it gets where it needs to be, as quickly as possible.
The problem with SDN is that it’s a relatively unproven approach. Manufacturers are starting to look more closely at SDN, with OpenDaylight, a collaboration project set up under The Linux Foundation, moving to unite the industry and create an open and transparent approach to SDN. Reports show the group is making progress on developing this common framework, but in the grand scheme of things, there remain few examples of SDN in practice.
Probably, the most well-known implementation is Google’s OpenFlow. Good news, if you’re utilizing Google’s massive network, where its servers determine from the outset, is that your searches and Gmail will be prioritized before anything else. Not so great for the other enterprises running across the Google network.
Data efficiency doesn’t mean cost efficiency
Another problem with SDN is that it’s an approach that is typically out of reach of the enterprise. Most solutions are targeted at large data centers and there are resource constraints and hefty price tags that only serve to put these solutions out of reach of the typical enterprise.
SDN is also not about efficiencies in terms of cost, but efficiency of getting data where you want it to be right away. SDN switches typically cost five times more and this price won’t come down until uptake increases. But greater adoption won’t happen with such a hefty price tag on a relatively unproven approach. It’s a catch 22. Perhaps if you’re backing Google then the benefits are there, but what if we are doing it the best way already? The enterprise could be left high and dry after making a significant investment.
Implementations of SDN are not straightforward either. Standard switches are no longer adequate, with entirely new software required to enable an SDN switch within the network. It’s certainly not an ‘off the shelf’ solution, and this becomes difficult for the enterprise.
Driving down operational costs
It may well be too early to look at SDN for majority of the enterprises, but managing the network effectively is critical now. There is a clear requirement to drive down costs and reduce network complexity and that means putting the focus on practical aspects of managing and upgrading a network.
Centralized management of the network and the effective integration of network management systems should be the short-term goal for today’s IT manager. This will enable new switches and network additions to be implemented and configured with ease, from a single place.
It’s a move that will support future SDN adoption, as the technology becomes more proven and cost-effective, ensuring that IT departments can transition to this emerging framework with ease.
I’ve little doubt that SDN has the promise to become the future of network management, as not only data but switches are programmed from a server. But until then, IT managers have a fundamental role to ensure the network today is working. Create a resilient and efficient network now that can be managed centrally, and use this as a solid base to integrate SDN in the future.