How SDN Will Help Network Engineers

  • Rating: 
    0 votes
    Vote up!
    Vote down!

At Cisco Live this year, the company's software-defined networking platform, ACI, was the hottest conversation topic. While the idea of SDN may be intimidating, Patrick Hubbard, Head Geek at SolarWinds, explains how it will ultimately make network engineers' jobs easier.


pace of change

Networking jobs will no doubt undergo radical changes with SDN, but I wonder if the pace will be the five years Patrick predicts. It seems like actual SDN adoption still is very slow and that there will be plenty of legacy networking infrastructure around for quite a awhile.

Re: pace of change

I was talking with the company Packet Design, and they are using SDN in very cool ways to enable their route analytics platform. It can integrate traffic performance and predictive analytics o manage resources over very large and carrier networks. It's in very early development -- so as you note, definitely not affecting your average network engineer yet. But that type of visibility should transform network management.

Re: pace of change

Agreed.  I don't see wholesale changeover to SDN in five years any more than the adoption of IPv6, (although SDN is going to make that easier too).  I'm looking at it from the perspective of where we were when I set up the first ESX server in a datacenter, to where that datacenter was five years later.  The most notable change was five years later you couldn't get a job in the datacenter without VM admin skills. There's lots of complexity in SDN adoption, and we'll have both conventional and policy based configuration side-by-side. 

Re: pace of change

One of the issues with talking about SDN is defining SDN.  I've seen a bit of disagreement on what the definition of SDN. 

If one were to take microsoft's definition, then anyone using virtual machines is already using SDN since virtual machines rely on virtual switches which can be configured with a number of options that mirror traditional switches.

Re: pace of change

@MarciaNWC> I think 5 years is entirely reasonable for some percentage of the networking population. There will then also be a large percentage that has not adopted "SDN" (however we choose define that) and pointy-clicky interfaces. Even if every vendor is selling pointy clicky, there will still be a large and slowly-aging population of equipment that isn't or can't be supported, or users who won't pay to add this functionality. Think of it like IPv6 in some ways - it's not a "this or that" adoption; it will be a more subtle migration over time until the point where to use the old technology simple isn't viable. Though let's hope it's a touch faster than IPv6, eh? 

Re: pace of change

You raise an interesting comparison with IPv6, and I enjoyed your recent blog post on  IPv6 and SDN adoption.

Re: pace of change

Just like computers in the 80's, the SDN technology looks promising. However, it is still too early to know how Software Defined Networks will turn out in the future. One thing is for sure, SDN will redefine how we manage networks and it is made possible by looking into the basic building blocks of our networks and tailor-fitting it to the networking needs and trends of the present and for the future.

Declarative Configuration

The key to SDN - and the changes that were mentioned in the video - is the move from procedural configuration to declarative configuration. Ultimately it looks like we will finally be able define WHAT we want to do, and worry less about HOW to implement that in the network on a box-by-box basis. That's the huge shift of SDN, however you choose to define what SDN actually is, and whether you want to include overlays as well as underlays in the picture. Pretty much all the tools and protocols cropping up in the SDN ecosystem are in support of that goal, I think.