CEOs can come across as the chief sales officers in interviews, aggressively repeating bullet points that may or may not answer the question at hand. Paul Maritz, CEO of VMware, is not that type of guy. From all appearances, he is a deep thinker with a philosophical bent -- not to mention he has bona fide technical credentials that stretch back to first half of the 1980s, when he helped Intel create programming tools for its emerging x86 platform.
Before becoming CEO of VMware, where he replaced founder and CEO Diane Green in 2008, Maritz was best known for his 14-year stint at Microsoft, during which time he oversaw the development and marketing of Windows 95, Windows NT, Windows 2000, Visual Studio, and SQL Server, as well as the Office and Exchange product lines.
Clearly, Maritz has been eyewitness to tremendous change in the industry, which he sees ramping up again as we enter the post-PC era. At VMware, Maritz has helped his company solidify a dominant position in virtualization, but as he made clear in a recent interview conducted by IDG Enterprise Chief Content Officer John Gallant and InfoWorld Editor in Chief Eric Knorr, he has no intention of standing still. The conversation began with Maritz's assessment of where VMware stands today and how it plans to step beyond virtualization.
John Gallant: People know VMware -- your strength in both server and desktop virtualization -- but what's your overall strategy going forward?
Paul Maritz: We're at an interesting juncture in our history. The company has obviously been very successful and continues to have a very compelling value proposition, which is kind of remediating the sins of the client-server generation. So people have bought our technologies to deal with server sprawl and have saved a lot of capital expense by consolidating servers -- and in the process found that they can actually address some operational efficiencies as well. And that's what has really propelled our company to date.
It's a great position to be in. But we're also cognizant of the fact that, unless we think ahead, we will be seen as the closing chapter of the client-server generation instead of the opening chapter of the next generation. We believe there are fundamental shifts happening in the space at the moment. And having been very successful, we've -- for better or for worse -- managed to attract the attention of some very well-funded and entrenched competitors. We have a target painted on our backs by them at this point.
This is not a time for us to be sitting on our laurels and congratulating ourselves. We have to gird our loins for the next lap of the journey. And because we are a very successful, but still a medium-sized company -- and we have determined competitors -- we have to use change to our advantage. We have to ask: What are the fundamental things that are changing in the world, and how can we get aligned with those changes and use that to our advantage?
Eric Knorr: Tell us about the changes you see taking place.
Maritz: We have a sort of simplistic way of looking at the world, which is to say that we're not subscribers to the reverticalization of IT. We don't think that the profound thing that's going on in the cloud era is a "bigger box" theory of IT -- that you'll get a complete vertical stack from a single vendor. We think that the cloud, whether private or public, is actually the antithesis of that. The cloud is actually about a new horizontal stratification of IT -- and we need to align ourselves with that.
Those horizontal stratifications, simply speaking, are about a fundamental transformation of infrastructure into a new, more automated layer. A transformation of how applications are developed on top of that, and then finally, a transformation in how the results of applications are delivered to the end-user. We think profound change is happening in all three of those layers.
Gallant: You're known for one of those layers.
Maritz: Obviously, our business today and the vast bulk of our revenues come from the infrastructure layer. The good news in our favor is that virtualization as a technique to transform infrastructures is no longer something that we have to evangelize. That's well accepted, and you know, if you look at the Gartner hype cycle, I think virtualization is kind of out of the slough of despond, and we're kind of on this slope of enlightenment, just about to reach the plateau of productivity.
So if you believe Gartner's numbers, if you look at the number of x86-based server applications -- and this is sort of the real metric to use -- Gartner would say that in the world today 40 percent of the survey applications that could be virtualized are virtualized and are running on virtual infrastructure and predominately on VMware.
That number is growing at about 10 percent a year, which means that a substantial number of our customers are no longer using virtualization on the fringes of their data center but are really starting to use it as a central strategy -- not just for running tier 2 and 3 applications, but for tier 1 applications as well.
We refer to that as our "breadth play" and obviously we need to continue that -- we need to do everything that we can to enable it to continue. But then we're going to have to start executing a depth play, because eventually we'll run out of server applications to virtualize. So we need to allow people not only to virtualize but to do more things in the context of virtualization, really start to tackle issues relating to operational efficiency, resiliency, and security within the virtualized world.
Knorr: How does that strategy align with your current product offerings?
Maritz: So we're moving to say we have vSphere, which we are continuing to improve to enable it to host as many of the server applications in the world as it possibly can. In other words, let's get as close to 100 percent as possible; let's make sure we remove any technical roadblocks to getting to the 100 percent. We have a few customers who are at 100 percent, but most are still way short of that.
But then, in addition, we are building toward a suite of products that will allow us to execute the depth play. We want to allow them to not only increase the quantity of virtualization, but the quality of virtualization. And that's why we basically think of ourselves as building suite of five products at the infrastructure level.
The next one is vShield, which is really the framework into which virtualized edge functions plug. Because as you build out this pool of virtual capacity, all these servers and storage and networking elements are now linked together through virtualization. We're going to move things around within that pool to improve efficiency and resiliency, so we'll not only move things around to load balance, but we'll move things around too for availability reasons, etc.
The protective functions around an application -- the firewall, the load balance, the antivirus engine, the data loss engine -- they're going to have to move with it so that all of those functions today that are typically encoded in a physical device that's clamped onto a wire somewhere have to be freed from that wire and have to move around in the pool as well. We need a framework for that to enable that to happen.
That's vShield, and we're working with all the usual suspects in the industry to get them to turn their functions from physical appliances into, figuratively speaking, virtual appliances that can plug into that framework: virtual load balances, firewalls, routers, antivirus engines, etc.
Gallant: So a framework for virtual appliances.
Maritz: The third element you can think of -- and I use this term very loosely -- is the user interface to that infrastructure. In other words, how does the producer of that infrastructure expose it to his customers? How does he make it easy for people to come, on the one hand, to provision the applications into that pool, associate policy with those applications into how they should run, and then give them metrics back as to how they're performing and consuming infrastructure?
That is what we call vCloud Director, which is the layer that adds the user interface, where you can basically describe your workloads, associate policy to them, and get metrics reported back as to how much infrastructure they're consuming, and so on, which allows an internal IT organization to start behaving more like an internal service provider to internal customers.
The fourth element is our Site Recovery Manager. It's a suite of software that addresses cross-site disaster recovery, because that's one of the common uses for this new infrastructure, where you have two active sites backing each other, etc. Site Recovery Manager provides the layer of functionality in terms of allowing you to have cross-site continuity.
Knorr: So vSphere, vShield, vCloud Director, and Site Recovery Manager. What's the fifth part?
Maritz: The fifth one is actually around the monitoring and management of the big pool that you've created. It's an interesting one because it speaks to this issue of a new horizontal stratification of IT that I was talking about.
Imagine trying to do this on a vertical silo basis. In other words, you try and find the app and then you manage all the way from the app down to the hardware underneath it. Now when you create that pool you're essentially cutting that stack right through the middle. And the pool provider on the one hand is going to know less about the apps, and the apps, of necessity, know less about what's going on inside the pool.
If you're the custodian of that pool, you can't look at any one individual app and know that your pool is healthy. What's more, if you turned on all of the logging capability of all the elements down in that pool, it throws so much information at you, that if you've got hundreds of servers and hundreds of storage areas there, you'll never be able to make head or tail of the information coming out of it.
You need to look at a different way of managing and monitoring in this cloud-based horizontal stratification of the world, so we've invested in a product called the vCenter Operations suite, anchored on a technology that we acquired last year, which takes a statistical analytic approach to how you monitor and manage infrastructure.
Gallant: How does that help customers?
Maritz: What it says is you're the end-user, don't try and figure out which of these logs is really important, which is not. Just give it all to us, just give us absolutely everything you've got, we'll [manage] it, we'll build a statistical model of your infrastructure and we'll tell you when we think your giant pool is going outside of the bounds of what we've historically seen as normal.
It's a statistical model, so every now and then it'll give you false positives. In other words, it'll say, We think you've gone outside the bounds of normal. And you'll say, Don't worry, it's just the end of the month. And it will learn from that.
But what we've found it that model tends to be much more sensitive than humans to important changes in the environment. It will start to detect that drift outside of normal before a human will detect it. It will say, Look, here are the five things that have driven the model outside of normal, we think you need to go look at these five things.
Knorr: You might say that these five elements together provide management of the private cloud.
Maritz: Yeah. We're deliberately very precisely trying to build up a cloud infrastructure suite, and I'll explain how we see it being applied in a hybrid cloud model. We are not subscribers to the belief that all of the world's computing is going to be either internal or external.
Knorr: Yes, most customers we talk to say they're most interested in the private cloud. So how did customers inform your vision of the cloud? What are they telling you that they want?
Maritz: What they said to us in essence is: We would like to make business decisions rather than technology decisions about when we run things internally or externally. And we'd like to be able to change our mind about that, so we'd like to be able to take something out of the external cloud and if we don't like it we'd like to be able to take it back, and vice versa. And we want to deal with a set of service providers who are willing to talk enterprise language.
You know, a guy who says "we take no liabilities" isn't someone you can do substantial business with over time. If we're going to really do this, we have to have people who are willing to engage in enterprise quality. We actually have about 1,000 service providers today who license vSphere from us and are operating some kind of service. We know that because we have an alternative licensing model for service providers, where they can essentially rent our software rather than paying up front for it. We know that very accurately.
Knorr: Many cloud service providers, of course, are going with open source virtualization solutions.
Maritz: We have a lot [of providers] who like the quality of our software, and they like it because it's familiar to what people have internally. So we have about 1,000 folks who are operating all sorts of businesses on a service provider model.