Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

SDN user group focuses on use cases

John Dix | Nov. 11, 2013
Nick Lippis talks about what he has learnt about SDN from the recent Open Networking User Group in New York City.

So you posted a number of potential use cases and attendees got to vote for the ones they thought were most pressing/important. What did you find?
The top three were integrated Layer 4-7 network services, virtual network overlays and branch office wide area networks. And all three have to do with automation. 

Regarding the first one, integrated Layer 4-7 network services, the only reason we have appliances, load balancing and VPNs and firewalls and IPSs is to make up for the inadequacies of the TCP/IP protocol. So you have all these separate boxes, separate management systems and separate vendors you have to contend with. And frankly, IT doesn't want any of it. They're done with appliances. They want the functions integrated into an overarching overlay strategy so when you fire up a workload you can easily add the various network services you need, whether it's load balancing or firewalling, and you have one common management system.

We don't want to deal with how to chain appliances together. We want them integrated and we want to get rid of all of the operational burden that's required to manage and maintain them. That's what the first one is really all about.

OK. And the second?
It's all about open overlays and choice. A good example of that is what OpenStack is offering with Neutron, where the modular Layer 2 plug-in can be anything. It can be VXLAN, it can be GRE, it can be STP (even though that's proprietary). You use the underlay, the signaling and the routing you have, to support a service that's overlaid on top of that. 

And the service that's been talking a lot about of late has been Virtual Machine-to-Virtual Machine, but there's the optical overlay and there's also the wide area network overlay as well. So what this use case says is that, "Overlays are good. We want them. We want them to be open, and we want choice in how we deploy them."

How does that differ from the vision that VMware puts forward with its Nicira technology?
That's primarily focused on the VM-to-VM piece, and that's one kind of closed monolithic stack. There's nothing that's open there. You can't swap software modules into and out of that stack. 

And the third one is about branch office support?
Yeah. At the branch office what the industry has done is stack up appliances for everything from unified communications to WAN optimization, firewalls, routers, wireless network controllers, etc., so if you have 10,000 offices and four appliances per, you have 40,000 appliances spread all over the place. So what this one says is, "We want to integrate all those branch appliances in software and be able to control the bandwidth to those branch offices a lot more effectively and efficiently. I was pleasantly surprised to see that become one of the top three, because it's a big issue. It's a huge amount of cost, and it's a huge amount of operational burden as well.


Previous Page  1  2  3  Next Page 

Sign up for CIO Asia eNewsletters.