If anyone has not listened to the PackPusher Show 112 – SDN’s Potential as a Displacement Technology With Jeff Doyle & Bill it is a must listen. A comment made by Greg who has rationalized and articulated the SDN conversation as well or better than anyone so far got me thinking. It was something I had never heard anyone else say except for vendors.
“The message we keep getting back from the vendors is they can’t identify the use cases for SDN”, “…The vendors aren’t yet willing to bet their money down on a piece of software that uses SDN and thats where we are at is waiting on some vendor to show some leadership to build it.”- Greg Ferro
That sums up about every vendor I have interacted with around SDN other than the couple two/three VCs. The SDN use cases for anyone who has ever been deeply involved in networks and has a basic understanding of the science of computing should be able to roll a list a mile long out. Yet vendors keep giving us the line of well everyone says we can do all of this with todays technologies, really?
This quick retort of customers not demanding SDN, is obviously not just a tactic to slow the demand down (hopefully) but is just as much execs, lawyers and marketing grappling with something as out of control as SDN and orchestration and working to monetize the rapidly evolving computing frameworks. There are evangelist inside all of the companies fighting the good fight of evolution and those are the ones we will never hear about buried in the basement of building Z and I salute them. Companies are fortunately taking risks, albeit slow and calculated. The excuse of not hearing customer demand is wearing thin and those would be wise to figure that out sooner than later.
The Northbound application for the data center is already here, it’s called Quantum the network API that sits between the data center orchestrator and the network resources. As with every other example of computing silos the networking side of the house is bringing up the rear. There is no reason that doesn’t go well beyond the data center. If everything is decomposed to just another resource to consume we can begin focusing on neglected and broken policy to serve the business objectives rather than the IT objectives.
I think the hidden gem under the hood in Greg’s statement is that vendors can no longer get away with playing dumb. 10 years ago a vendor could stifle innovation by isolating customer requests as niche and continue pushing their own best interest down the customers throat. The world has changed and commoditization of hardware has unshackled the consumer to have the freedom to execute what works for their architectures and business strategies. The mobilized community of practice has just begun. The rules will be set by the community leadership and business needs to scale infrastructures with logarithmic financial expenditures.
The roadmap moving forward is as simple as looking at the evolution of the x86 market. Commoditize the hardware, decouple the OS and build proper abstraction layers and primitives. With the best of the best from hyper scale consumers to the thought leaders like my brilliant friends in the fledgling networking community it is not a question of if but when. Can we learn from the x86 evolution and trim the time of getting networking closer to par? Next time a vendor or a colleague tries to defend an archaic status quo throw some links at them to RTFM.
