Thursday, March 21, 2013

What is SDN Good For?

At least in some quarters, we are in a growing hype cycle related to software defined networks. As I simplistically understand it, SDN is about separating the "control plane" from the "data plane." 

In a probably over-simplistic way, that means putting intelligence in a headend or central office location, while using much-simpler, cheaper devices "out in the network" that only have to deal with "data," while the centralized controller does the computation that formerly was "out in the network."

The broad analogy will be familiar to those who once worked with "client-server" computing architectures. It isn't a case of "mainframe-dumb terminal" models, because the SDN clients are not "dumb" devices. But the big computation loads are handled centrally (control plane) while the many distributed data plane devices can work fast because they don't have to deal with local computing chores, and just handle the data.

As we typically do, there will be years of wrangling over "what SDN" really is, and how it is implemented. There will be many attempts to graft SDN onto the existing base of platforms that actually have intelligent devices scattered all over the network.

What probably will be lost is the "so what" part of the discussion, for service providers or their customers. 

Most lost in the discussion will be the end user benefit. But some might argue that's what we really should be talking about. 

It is perhaps not required that SDN be used to provide a user-defined set of policies relating to use of bandwidth. That could logically be provided other ways, including the use of intelligent devices scattered throughout the network.

But it might arguably be much easier, much faster, more dynamic and more affordable to supply end-user define policies on an SDN network than it is today. 

For service providers the advantages might be even simpler. SDN might enable "bandwidth on demand" in a reliable, robust, affordable way. 

But all that will likely be lost in the wave of hype we are about to start hearing. 

No comments:

Will AI Actually Boost Productivity and Consumer Demand? Maybe Not

A recent report by PwC suggests artificial intelligence will generate $15.7 trillion in economic impact to 2030. Most of us, reading, seein...