- Marc Goodman
When I was a kid, my friends and I used to swim in a neighbor’s pool every day in summer, and we would compete to see who could hold their breath the longest under water. I was pretty good, but in the back of my mind, I was always concerned about making it to the surface before my air ran out.
When I think about new technologies like software defined networking (SDN) that have the potential to cause sweeping new network advances, I think back to inhaling as much air as I could before immersing my head. With all of the potential problems that could occur with SDN implementation, you’d better take a really deep breath before diving in.
The promise of SDN WAN
When SDN first began to dominate conversation in the networking market, most thought of it as a data center technology. The idea was to decouple the control plane from the underlying physical network and then use a centralized controller to manage the entire data center network as one. This centralized controller would offer the ability to program specific flows between nodes and eventually enable network virtualization where software network instances could be spun up on demand.
But soon SDN research turned to the WAN. After all, why not use the same architecture to manage the networks between geographically dispersed data centers and offices? We could even have the ability to spin up virtual networks over long distances that could cross network domains.
Going further, SDN could improve WAN performance and flexibility. WAN optimization vendor Silver Peak envisions enterprises and cloud operators using hypervisors to allow non-networking employees to directly manipulate and provision networking infrastructure to support their applications. For example, an employee from marketing, sales or finance could adapt network, storage and compute resources through a simple user interface to support a replication process.
Meanwhile application delivery vendor F5 sees SDN being used to manage Layer 7 networking services and to ensure security, acceleration, optimization and routing in the WAN.
Lots of potential, but even more unanswered questions
All of this is exciting, but the problems along the way will be plentiful.
For example, WAN disruptions could occur when adding a new software layer to create a virtual network that is independent from your physical network. Consider the potential management challenges that could arise from no longer configuring each WAN hardware device, but instead programming them all with centrally managed software in a controller. It sounds great, but will this actually ease network management or make it more difficult? What’s more, can one controller handle an entire WAN? If not, how will we manage an environment that must have multiple controllers? Will there be a controller of controllers?
Then there is the problem that multiple SDN architectures are emerging and it’s increasingly difficult to know where to invest.
While the Open Networking Foundation is developing the OpenFlow protocol and the OpenDaylight consortium is expected to work on northbound application standards, some vendors in the meantime are developing proprietary strategies.
That can be a real problem when it comes to the WAN. What if an enterprise has deployed one SDN solution but later needs to integrate its WAN with a cloud provider that is using a different SDN variant? With all this uncertainty, it’s hard to feel confident in spending money on SDN-compatible network devices.
I’m not saying SDN won’t eventually be worth the investment. I simply think it’s a good idea to take a deep breath before you take the plunge, and evaluate how SDN might impact your business in the near and long-term. Will it allow you to break the surface victorious, or leave you gasping for air?
About the author:
Marc Goodman is a marketing consultant with over 30 years’ experience as a marketing professional in the technology industry. He has a successful history of building leading brands for emerging companies, managing corporate and product marketing strategy, and working in the trenches on tactical program implementations.
Is SDN washing the new cloud washing?
SDN and OpenFlow, not the same thing!
Northbound OpenFlow applications on the way: Watch out Cisco!