VMware's Chief Architect: Don't Be Scared Of SDN, Be Prepared4:00 PM EST Thu. May. 09, 2013
If anyone can speak to the impact software-defined networking will have on the IT landscape, it's Martin Casado, chief architect of networking at VMware. Named a top innovator of 2012 by CRN, Casado is the founder of Nicira, the SDN startup VMware acquired for a whopping $1.25 billion. He is widely considered one of the founding fathers of SDN and the concept of virtualized networks.
That's why CRN joined Casado at Interop to hear his take, first-hand, on one of the industry's hottest trends. After participating in a panel Wednesday with other industry executives trailblazing the SDN space -- including Rajeev Nagar, group program manager, Windows Core Networking at Microsoft, and Rajiv Ramaswami, executive vice president and general manager, Infrastructure and Networking Group at Broadcom -- Casado spoke with CRN about everything from the impact SDN will have on the channel to why VMware and Microsoft may end up seeing eye to eye after all.
Take a look.
I honestly think we could have walked onto the panel, and then walked off and didn't say anything, and the primary point still would have been made. And the primary point is that the voices of networking are changing. You've got [Microsoft and VMware], which traditionally haven't seen eye to eye ... and then you've got [those] server software vendors, two of the largest in the world, talking to a server silicon vendor. And the dialogue is the same.
I think it really points to a significant trend when the players shift in this way. For me, that was the primary take-away. So one take-away is just the people on the stage, and just the alignment and the agreement. That alignment is that both Microsoft and [VMware], according to the panel anyway, agree that ... the software players are building more networking functionality. And we totally agree. Before the panel, I was talking to Rajeev [Nagar] from Microsoft, and we were agreeing about everything. There was nothing to disagree about. We know that we are moving in this direction, and it's so significant."
In networking, traditionally, when we think about supply chain, we think about boxes. We think about does [the gear] come from a traditional vendor, does it come from an OEM, does it come Frye's? I think the most impactful point, vis-a-vis the panel, is that there are more players and more channels, and it's not just about the hardware supply chain. So, for example, I provide software and that software has networking functionality. So it's a totally different supply chain because it's coming from software.
If you look at somebody who does their own development, like Microsoft, who does a lot of their own internal development for their own data centers, the supply chain actually becomes internal. So the box supply chain is going to have to respond to this horizontalization. And I think that we are all a little bit foolish if we try to predict what's going to happen with that. But we should all acknowledge that there are more suppliers and a different channel now, with different functionality and, in particular, software.
I believe the network sale is about to change. And [the channel] is probably the most important piece of the network sale. It's a very important piece. And what that means is that [solution providers] are going to have to start a dialogue -- and some of these are already ongoing -- but the dialogue needs to broaden to not just traditional box vendors but to the software [players], and understanding that.
We already see this starting to happen. But, really, if [solution providers] read this and get one thing to take away, it's 'I should go talk to VMware. I should talk to Microsoft and I should understand what this means to me.' And it just means there are new relationships that need to be formed. It's not some cataclysmic, end-of-the-world event. It's really just a shift in relationships.
I don't think anybody can really agree on what SDN means. And I certainly don't know what it means anymore. I just want to be clear about that. But I think there's a sense that there's change in the air. And that change is around horizontalization and that change is around software-consuming functionality. We can call it whatever we want. To me, this is the realization of network virtualization.
I think people know the networking world is changing, and it has everybody on their heels because nobody knows which way the world is going to converge. So I think everybody is threatened, but not really threatened by SDN. They're threatened by the knowledge that there's change. And I think it requires redefining roles, and I just think it's scary. ... It's not as as simple as 'such and such is going to be made irrelevant.' It's really about redefining roles. It's like, what is the role of the traditional networking vendor if the functionality is being provided from software, what is the role of the channel, what is the role of the network architect? It's not obvious what the disruption is going to be.
In my opinion, network architects, with their existing skill sets, continue to be incredibly relevant. There is still a lot of networking to be done. But I think, if you ask that question, many people give different answers.
Now, if [network architects] want to broaden their purview, they need to understand the virtualization layer. There's an intersection of layers. So the more they can understand virtualization and virtual networking and how it works, I think the more broadly they can apply their expertise. But it's not like networking knowledge goes away, and I think there's a misconception about that.
I do think we need to redefine [network] operations, which is not an architect, it's operations. Right now, we do a lot of [network operations] manually. And if you automate it, you no longer have to do it manually. But that doesn't mean people's jobs go away. It just means it changes. Just like what happened with server virtualization; you don't need to rack and stack as much, but you need to understand how to deploy and manage and troubleshoot the software.
So I think there is a shifting of responsibility and roles. But these are all going to be redefined. And, to your earlier point about who's nervous, everyone is nervous.
I think compute virtualization snuck in in the still of the night. People did not understand the magnitude of the change. VMware pretty much created this industry. And the initial pitch was server consolidation, which is a very simple pitch. It's 'instead of having two servers, you can have one.'
But what it ended up doing is transforming the way we actually manage and operate computers, and then saved tens of billions of dollars a year for customers. So it snuck in in the still of the night and then it changed the world.
So I think everybody realizes the potential of network virtualization. I think when it comes out, it's going to come out loud. It's not sneaking up on anybody. And I think it's a good thing. It's going to come quicker and it's going to hit a lot harder.
If you had asked this question three years ago, I think the common answer would have been service providers or mega data centers, but I don't believe that anymore. Any company can be an early adopter. A financial company can be an early adopter, a consumer company can be an early adopter, a service provider can be an early adopter.
So think of this as not being confined to a vertical, but being confined to a horizontal of early adopters. Those who want to innovate in their infrastructure will adopt this technology. It's not a vertical play. We have customers across all the verticals, as a case in point.