VMware CEO: What we plan for the post-PC era

Paul Maritz, CEO of VMware, sees virtualization as one piece of the platform VMware intends to offer a mobile, cloud-enabled world

1 2 3 Page 2
Page 2 of 3

Within that community, what we've elected to do is to try and pick a very small subset of them who are committed to have the same suite in their public clouds. In particular, that user interface is going to be common between the two. The way that you describe and secure and manage your workloads will look the same internally versus externally. What's more, there are people who are not only willing to talk enterprise, if you like, but they have the means to back it up.

In other words, it doesn't help for a vendor to say, We're willing to accept liability. The question is, if you ever had to cash that check, is there something to really go sue or not? So we've created this partnership that we call the vCloud Data Center Partnership and try to recruit people in that who will look like credible enterprise service provider partners. Today we have Verizon and Terremark.

Knorr: Right. Those are some reliable partners.

Maritz: At least there's something to sue there if they go wrong: Verizon, SingTel in Asia, Colt in Europe. Colt is the City of London telephone company that carries all the fiber-optic cable for the financial industry in Europe. Or Softbank in Japan, which is the second largest telecom operator in Japan.

Those are the four anchor tenants, and there's Bluelock, a small guy that we put in there just to keep them honest and show them that things can actually get done in as short of time scale as [they're] accustomed to. We'll expand that by probably another three or four over the next several months.

Interestingly, those companies have started to discuss among themselves forming a consortium so that somebody can say, I want to deal with one entity that can give me coverage in Europe and Asia. Because there's always going to be regulatory reasons why certain things have to be done in certain jurisdictions independent of technology. How do I work with somebody who says, If the Singapore government requires me to do certain things in Singapore, I can do that, or Japan, or whatever?

Knorr: And how does this elite group of service providers relate to your enterprise customers?

Maritz: Well, that's hybrid cloud, so we're saying to them: We think most of our customers are going to go ahead and implement private clouds. When and if you're ready, there will be this community ecosystem of external service providers. And it's important that it's an ecosystem, so you're not beholden to any one of them, they'll keep each other honest, who will be ready when you're ready, to allow you to make business decisions about whether you want to take an application or set of applications and run those in an external cloud. And if you don't like it, take it back again and move it somewhere else.

Knorr: Is there a direct relationship between this and Vblock?

Maritz: No. A lot of those guys are, but not exclusively, using Vblock. Vblock is really a lower-level construct. Vblock is just an efficient way of providing the hardware basically.

Knorr: You're saying then that they'll be using all five elements of your stack, of your infrastructure, so they could appear as a transparent extension to my enterprise infrastructure.

Maritz: That's exactly what we're trying to do.

Knorr: Wow. OK.

Maritz: Right now it's "your mileage may vary," but that set of six that I mentioned are committed to have 22 data centers started up by the end of this year around the world -- a substantial investment.

Gallant: So, Paul, once you go beyond the enterprise infrastructure play into areas like platform as a service and beyond, how do those elements play out?

Maritz: I'll talk about that. Those five elements I talked about make up one layer. We think of it as three layers.

Layer 1 is essentially our take on infrastructure transformation, which is very firmly in the context of this hybrid cloud -- that most businesses are going to have both internal and external infrastructure. Both need to become more efficient in a fundamental way, and that infrastructure has to handle existing applications above all else.

Customers can't afford to rewrite or abandon their apps, so that infrastructure has to be good at handling existing applications. That's why virtualization is so important, because the real profound thing about virtualization is it really encapsulates an existing app. When you virtualize, you're taking an application and the operating system and the middleware, you're putting in a black box, cutting the tentacles of complexity that tie it to the details of the underlying infrastructure -- which allows you to kind of jack the black box up, slide the new functionality in, and slide the black box around.

Gallant: All at the infrastructure level.

Maritz: Virtualization is kind of the on-ramp for existing apps on that journey. Now, the challenge is it's not the only thing that'll happen going forward. Customers are now very much on an infrastructure transformation journey; they've invested in this over the last two years, they've by and large seen good returns.

They know they have more to do, but they now are starting to say the fundamental issue that we face is really our applications. Because if we're stuck on 20- and 30-year-old application code that was written for an era of paper bills, we're going to have a problem servicing the Facebook generation. People are going to want to see and consume and manipulate information in a far more flexible and fluid way than they've done in the past.

We believe that over the next 5 to 10 years, as big as infrastructure transformation is, that there's going to emerge a fundamental application transformation that needs to happen. Given our theme of trying to look for the important tides of history, we see that developers over the last 5 to 8 years have kind of revolted against complexity and have taken matters into their own hands. You see coming out of the development space all of these new modern programming frameworks, whether it be Ruby on Rails, Spring, or Django, none of which came from the established vendors.

There's this developers' old saying: A pox on your complexity, we're going to do this in a simpler way. That's certainly where the new lines of code are being written, so that's one fundamental change. The other is that developers increasingly want to consume things at a high level. They don't want to have to worry about cobbling together middleware and mapping into virtual machines, etc. They want all that just handled for them.

Knorr: That is certainly an ongoing trend.

Maritz: And thirdly, the new data fabrics, because coming out of the consumer space, they've already had to confront huge amounts of data which can't be handled through a traditional relational database model. You've seen the Hadoops and Cassandras and Reax, and they're all coming out of the consumer, Internet side of things.

We think that new programming frameworks wanting to consume things at a higher level and new data fabrics are really a disruptive change in that industry -- and they will be the foundation on what we think a lot of application renewal will be done. We've started to invest in that space in the context of originally acquiring the Spring programming framework, which was technically, arguably the best of breed in this area. In trying to identify the next-generation middleware, we've acquired the GemFire distributed data fabric.

But in addition, it was our belief starting about two years ago that it would be an unnatural situation if the world ended up with a couple of highly proprietary über clouds. That would be going back to the mainframe era of the '60s and '70s, you know, where you basically could write an app for IBM's world or Digital's world or whatever. Moving between them was very, very difficult.

Then Unix came along as a way of kind of bridging that, and it got morphed into Linux and strengthened as it became an open source movement. It is our belief that the developers and the open source world in particular would not eventually tolerate a world of highly proprietary clouds, that eventually out of the open source world they would find a way to basically come up with a new cloaking layer that isolated applications to a greater or lesser degree from the details underneath.

The very loose analogy that you have to be careful about is, if virtualized infrastructure is becoming the new hardware, what's going to be the new operating system? What will play the role of Linux in terms of isolating you from any details down there?

So a couple years ago we hired two senior developers out of Google who had worked on Google's internal infrastructure, and one had a Microsoft background. I knew him as Mark Lucovsky, who Steve Ballmer apocryphally threw a chair at when he told him he was quitting to go to Google. The other was Derek Collison, who was a very senior guy out of TIBCO. He basically was the architect of the TIBCO message bus before going to Google.

We said, If you were to do this again, how would you do it? And in particular how would you construct a layer that filled that vacuum? That's what became Cloud Foundry. And we said if we're going to come up to that layer and if it's to fill that vacuum, we can't keep it proprietary, we have to make it open -- open source. That's why we not only released it but released it under an Apache 2 license, so it truly is an open layer in that sense.

Knorr: Do you see it as competing with OpenStack?

Maritz: It really sits on top of OpenStack, at least as OpenStack is currently conceived.

Knorr: Which is an infrastructure play.

Maritz: Yes. And we see Cloud Foundry, where we had to say you either play here or you don't play, and if you play you've got to be willing to see this layer go everywhere. And we're expecting it to go everywhere.

Knorr: Including to OpenStack providers.

Maritz: Including on OpenStack, including on Amazon. People are already taking it and doing that.

Knorr: In a lot of the coverage of it people were comparing it to Azure. Are they apples and oranges? It sounds like it based on what you're talking about.

Maritz: Basically, yes. Azure is a complete proprietary stack. I mean, in some senses, you can say they are both platform as a service, but we actually think there's a whole bunch of features that Azure doesn't have. It's conceptualized in a fundamentally different way. We have cloudfoundry.org, which is the open source project, and we've started up cloudfoundry.com, which is where a developer can go and start kicking the tires and actually writing an app in this environment and seeing how it works.

But there is this dual element to it. One is that it's actually Linux with a cloud, which is what we're trying to do there. And then we've said, OK, let's stand up an instance, of which we think there will be many instances, where people can go and familiarize themselves with it. We've been actually kind of blown away by the response. We have over 25,000 developers signed up in the first month.

Gallant: That's pretty amazing.

Maritz: To kick the tires ... most of them are just signing up to write the Hello World app and see how it works, etc. But still, it surprised us.

Knorr: It's a virtual circle in the sense that all the work done in the open source part comes back in and benefits...

Maritz: Well, it's actually an interesting layer, because it has three interfaces on it. On the top it has a set of formalized interfaces that you plug these modern programming frameworks into. So we announced it out of the gate with the plug-in for Spring, the Ruby family, Rails, Sinatra, etc., and now .JS. People have already started to extend that, so we have now got Erlang plugged into it and we're expecting others to get plugged into it as well.

It has a set of interfaces on the side, kind of figuratively speaking, where you plug in services. That's where you can plug in different databases, queues, etc. We came out of the gate with the usual suspects, you know, MySQL, Mongo, etc. But we expect people to plug other stuff in at the side. Then it has a binding layer on the bottom that binds it to the particular infrastructure that it's sitting on top of. Those are formalized, well-defined interfaces.

Gallant: Can I just ask one quick question on this theme? How do you monetize it?

Maritz: Two ways. One is that we think over time we can do hardened versions of this for the private cloud world, where that will be a classic open source-like model where you're basically selling service and support. We think that we can plug services into the side, so we do have some very valuable services there.

If you look at GemFire, which is a very interesting kind of sleeper in the world of middleware, it's one of the most strategic apps in the world -- the app that the U.S. Department of the Defense uses to track its assets, real-time assets, is built on GemFire. So all of their vehicles that have telemetry on them are reporting their status into GemFire. In theory, you can put up a map of the world and drill down into an aircraft somewhere and find out in real time how much fuel it has onboard as it's flying. That thing gets 60,000 position updates a second, which you can't do on a relational database. It just cannot be done. Those kinds of things we can plug into Cloud Foundry and monetize it that way -- very important.

Gallant: So how does that work in a service model? As a subscription or...

Maritz: Every service will have its own different kind of model associated with it, depending upon what makes sense in that area. So that's sort of what we're trying to do in the pass layer, which is to say -- how do we attach ourselves to the new programming frameworks? How do we attach ourselves to how developers want to consume infrastructure? And how do we plug new services into that? And deliberately try to shoot ahead there.

Knorr: Platform as a service has been the slowest to take off of all the cloud categories. What makes you think that Cloud Foundry is going to take off where others so far have not?

1 2 3 Page 2
Page 2 of 3