Wingspan

John.McHugh

Cutting Through the SDN Hype

by John.McHugh on ‎05-08-2012 11:51 AM (552 Views)

Brocade who was an early and unambiguous supporter of Open Networking standards, and in fact, publicly asserted our commitment to products which support Software Defined Networking (SDN) as early as June of 2010, has understood the value of this nascent trend for years.  Unfortunately, like most significant innovation cycles in the technology industry, the airwaves are getting polluted by the misinformed and overzealous. 

Let me attempt to shed some coherent light on some of the hype that I am hearing:

  1. “SDN is here today or will be ready to deploy shortly.”   If you think you could tune the ignition system of the new car you just purchased by reprogramming the fuel/air software on the car’s engine computer, then you are probably equally right on these two perceptions.  Conceptually, SDN is here.  OpenFlow demos have been running for years and products exist or will shortly exist to allow you to implement these architectures in production networks.  But that doesn’t mean you can or should.  The ecosystem of tools to let you understand what is going on in the network, program something better and then automatically and dynamically respond to changes is remedial if not non-existent.  Even if it did exist, or you had a crack team of network programming experts, the breadth of networking solutions to deploy is thin to say the least.  The first adopters will be companies who are in business to squeeze the most performance out of their networking infrastructure, and even they admit there is a lot of learning to do.  That is not most of us.
  2. “SDN promises to make the physical infrastructure irrelevant to the actual behavior of the traffic.”  This is the most popular myth and the current fodder for the perpetual perception that “Networking is commoditizing”.  Does anyone with an engineering degree really believe that someone could make software whose performance would be independent of the hardware underneath?  Anybody running Apple iOS 5 on an iPhone 3?  Separate yourselves from the hype and just think…the quality and performance of the hardware DOES matter and in fact may matter MORE in an open software model.  Anyone running their LINUX on a Commodore PC?  SDN may create separate markets for the software that controls the network and the physical infrastructure. But people who care about the stability, performance, supportability and cost effectiveness of their networks, will need to buy from network infrastructure companies who are committed, experienced and capable of supporting these mission critical ecosystems.
  3. “SDN is about to change the landscape of the networking industry.”  Probably not. Going back to the LINUX analog, did LINUX favor the white box manufacturers, the knock off companies and the hardware start-ups of the time?  No. Open software architectures punish those proprietary software and hardware providers who reject the value and concept of open software or those who attempt to deny its relevance to the industry.  Ultimately, a fully matured open software ecosystem runs on best-in-class hardware solutions from suppliers who understand how to create platforms to support high performance mission critical networking infrastructure. 

I’ll wrap up by assuring you that Brocade’s roadmap for devices which support SDN architectures is strong and will be growing.  Because Brocade is a company which focuses on networking architectures, we are not merely generating SDN solutions in a shortsighted attempt to replace existing traditional designs, but instead we are understanding how SDN enables, extends and enhances the new and current network use models which customers are deploying today. 

Comments
by (anon) on ‎05-16-2012 12:45 AM

Interesting take and IMHO a healthy "down to earth" statement. I'm glad I read nothing about "Let's try to do FCP with SDN"