Interview: VMware's Bogomil Balkansky Discusses vSphere 4

During a recent trip to the VMware campus in Palo Alto, California, I sat down for an interview with Bogomil Balkansky, VMware's vice president of product marketing. Our discussion covered the arrival of vSphere 4, VMware's market share in the SMB market, and what VMware's long-term strategy is for positioning their products against Microsoft's virtualization portfolio.

Jeff James: How much have you relied on your larger customers for feedback on product improvement \[for vSphere\]? Where does the inspiration for new product features come from--is it entirely driven by customer feedback, or is it based on the approach you want to take?

Bogomil Balkansky: It actually comes from two different directions, and we have a very interesting, very unique model of product development where on one hand, we do sort of the classical thing that most software companies do where we have a very structured process that is customer-driven--stay close to customers and track how they're using the software, generate ideas, and get customer feedback for how to improve things. So again, we have a very structured methodology. We have a customer advisory council, user groups, and multiple ways to source that feedback. And it gets filtered through and prioritized by product management, and then gets fed into our product development plan.

In parallel with that, we also have an engineering innovation track, where we give a free license, if you want, to engineers. There is time and it is encouraged for them to experiment with cool new things. Even if you don't know if there's any commercial applicability of something, like a crazy new idea.

Jeff James: That's similar to what Google does.

Bogomil Balkansky: Sort of. My understanding with Google is that it's very structured: 20 percent of your time. Here it's not really mandated to 20 percent of the time you have to do this, but it's very much encouraged and rewarded. Usually twice a year we have these big innovation fairs with R&D where engineers do demos of prototypes, or present the cool new stuff that they think they can get done. So that's the other major source of innovation. The two meet in the middle--here's what customers want, and here's the super cool new stuff that we think we can do that customers didn't know they wanted because they didn't know it was possible.

If you think about it, a lot of true, groundbreaking innovation wasn't really based on customer demand, I mean it was based on customer demand, but if you think about the cell phone, it wasn't invented by doing focus groups. Would customers love to do it? Yes. But if you went around doing focus groups for fixed phone service, you probably wouldn't have come up with the idea of wireless. That's not how it happened.

So sourcing ideas from these two fronts, they meet in the middle. And again, that's part of our product development process. Before we create a plan of record, we take ideas from both sides, prioritize them, and say "Hey, this is what we're going to do."

VMware Bogomil Balkansky

Jeff James: One of the new vSphere 4 features discussed during the announcement event was the new host profiles feature, which is aimed at helping enterprises that are managing large numbers of virtual machines. Maybe you can talk a about some of the other top 2-3 vSphere features that were requested by customers? And what is your own favorite feature of vSphere 4?

Bogomil Balkansky: I think one that has a lot of popularity with customers, for the same reason why host profiles are attractive, is the vNetwork Distributed Switch, which is this feature that we have enabled in our platform and Cisco has built a product to actually instantiate it. With the Distributed Switch, we have the ability to create or configure the virtual networking once for an entire cluster of servers. So instead of configuring a virtual switch on every host one-by-one, and again when you have a hundred hosts it gets cumbersome. It's a similar idea in many ways conceptually to host profiles. With host profiles you create one golden template, and you blast it out through every single host. With vNetwork Distributed Switch you take a whole cluster of servers, and in essence you configure the virtual networking once across the entire cluster, similarly very important to be able to manage that scale.

As we're getting into cloud computing, one of the paradigm shifts that is required is can you actually manage an environment of that scale. Instead of managing onesies and twosies, can you manage hundreds, thousands. So, both host profiles and Distributed Switch are the kind of features you need to have high-level management and scalable management.

Now, in terms of my personal favorites, I think fault tolerance is going to have a significant impact on the market, not only because technologically it's way cool. But you think about the amount of time and money that has been associated with providing that level of availability before--we're decreasing the cost and complexity of providing that level of sophistication of high availability by orders of magnitude. And I can't tell you exactly how many specifically, but I know it's multiple orders of magnitude. When you so dramatically change the economics of something, you tend to drive very interesting dynamics in terms of demand. So, think about the ingenious invention of Henry Ford's: he made cars much cheaper than before, wow, then kept expanding them out multiple times. And what we're doing with fault tolerance is that we can make that expansive capability so cheap, relatively, and so easy to use, so like in the demo today, virtually it's right click, turn of fault tolerance, click yes, and it's done.

I think people will really think differently in terms of what are the kinds of applications that actually deserve that level of application, because in the past it's been so difficult and so expensive that it's really reserved for the 0.1 percent of really important stuff. And there's a lot of disruption and downtime cost for less important stuff, but again a lot of people running around with nothing to do because the system is down, doesn't work. I mean, what do you do? "The network is down, let's go home." That's what people do.

So I think we'll see a significant expansion in terms of the number of applications or the kinds of applications that can really benefit from that level of availability.

Jeff James: With some of your new product editions \[VMware vSphere Essentials and VMware vSphere 4 Essentials Plus, both aimed at SMs with fewer than 20 physical servers\] it looks like you're trying to go after customers that VMware hasn't traditionally gone after. Is that a response to competitive pressure from Microsoft with Hyper-V entering the SMB market, or is it an answer to the current economic situation, or is it a combination of both?

Bogomil Balkansky: We've always gone after that segment of the market. Yeah, we love beating our chest \[about the fact that a\] hundred of the Fortune 100 companies are VMware customers, and 96 percent of the Fortune 1000 are VMware customers, and guess what--that leaves 129,990 customers who aren't. There's a ton of them. If you think about the classic stratification, yes: We have the global Fortune 2000, we have the more mid-sized companies like the tier below, and maybe we have 10,000 of those. The rest of them are SMBs. The overwhelming majority of our customers today are SMB customers. And they love what we do, because for the first time they have, in some ways you can call them toys, or tools that they have never had before, at a cost that, even the most expensive edition of our software is less expensive than most servers, only it allows one server to do the work of ten servers. So it has always been very cost effective even for small customers, and small customers honestly adopt even the most advanced features faster than the larger customers.

Jeff James: We have some readers that are doing that.

Bogomil Balkansky: I just remember, after the launch of VI3, I ended up personally emailing about 1,000 customers out of my mail account, and a lot of them were SMB customers, and I have so many wonderful things about "Yeah, I love DRS, because I'm the one guy who deals with IT here, and setting the system on autopilot, to load balance on its own, I've never had anything like that before."

And so, we're putting a lot of again, dramatically reducing the unit cost of computing. We've already made it accessible to small companies, and a lot of them are using it. We have over 100,000 small customers. So the challenge is there are so many small customers, it's hard to reach them all--how do you get to them? It's kind of a classic challenge for any frame. It's not that we don't have a valuable proposition that tremendously resonates; it's just how do you get to them. This is more of the challenge.

We're going into this from a tremendous position of strength. Microsoft obviously has tremendous install-base in SMB and has done tremendously well, but Hyper-V doesn't have much of an install base yet. They don't have 100,000 customers. Is it a defensive move? No. We have gone to SMB before, and of course, because we have so many more SMBs, we need to make sure that we continue to solidify our position in the market.

Jeff James:There has been some discussion in the media recently about Microsoft gaining ground with Hyper-V among SMB customers. Any comments on that? \[Note: IDC released some virtualization market share information in Q4 2008 that indicated Hyper-V was helping Microsoft gain ground against VMware in the SMB market. VMware has disputed some of the findings in the IDC report.\]

Bogomil Balkansky: We could definitely set you up with some customers. We also just completed our annual survey that we do. It goes out to VMware customers but also a blind market sample. There's some very interesting data to share. One of the very interesting dynamics is before Hyper-V finally hit the market late last year, for a couple years we were competing against a ghost. We were competing against this inflated expectation like "Ooh, when Microsoft comes to market, what is VMware going to do?" And having the actual product in the market has set the expectation right. OK, it's a product. Now we know exactly what it is. It's no longer this ghost, no longer sort of on our promise and exact feature list vs. this and that. And again I think the way the market has responded to it so far is, people are on the fence still about Hyper-V.

And I can appreciate why some people are getting up in this train of thought. This is the Microsoft playbook, right. Come, be a late follower on the market, wait for someone to blaze a trail, come out with an OK product, make it really cheap, bundle it with something that's already part of your cash cow, start with the bottom of the market, and kind of chip away, chip away. And I think because of that that's a very tempting train of thought to say "Oh, of course, it's the Microsoft playbook."

I would argue that, in many ways, we as a company are in some ironic ways almost playing by the same Microsoft playbook.

Jeff James: Do you think that has anything to do with Paul \[VMware President and CEO Paul Maritz, a former Microsoft executive\] now being at VMware?

Bogomil Balkansky: No, that's why it's ironic. What I mean by that is: what is Microsoft famous for? Come into the market at a very cheap price. If you think about what we are doing, again we are about transforming and dramatically decreasing the unit cost of computing, and giving people very sophisticated capabilities at a fraction of the total cost of ownership that was possible before. That's a very very popular value proposition and one that resonates with anybody. It doesn't matter if you are a Fortune Global 2000 or a small customer. I think it's very easy for present industry observers to kind of latch onto this train of thought, Microsoft is going to go to the bottom, because nobody actually talks to these businesses. Nobody has bothered to actually go out and asked a guy with three servers "What are you using?" Nobody has done it because while it's very easy to identify who the spokesperson for a Global Fortune 1000 company, it's very difficult to determine who the guy who takes care of Mike's Bike Shop servers is. You'd be surprised to find out that the sys admin at Mike's Bike Shop is using either our free product, ESXi, or he's using all the way up to VMware Infrastructure Enterprise.

Jeff James: We've recently ran some articles about cloud computing, and we heard from a number of readers about the topic. Some are already adopting virtualization and really looking forward to cloud computing, and others still have a lot of concern about security and how to manage identity in cloud, and asking questions about how AD will federate across on-premise and off-premise services. There's a broad spectrum of people who are exploring Cloud Computing. We've seen a lot of our readers who are small businesses owners be the first people to dive in and really take advantage of the latest technology.

Bogomil Balkansky: But again, there are so many of these small businesses. We have 100,000 of these customers, but there is still a ways to go. With our free products we have millions of downloads, with our original free product VMware Server and now ESXi. It's almost like the dark matter of virtualization, if you want. It's the uncounted kind of virtualization that is massive actually. When you see all these numbers of percentage of servers virtualized, most of these statistics only count paid product, but there are a lot of free downloads, so people are doing a lot more than is obvious.

Jeff James: We had talked before about how VMware really has a vision of eventually turning enterprise IT into a resource, like you plug your laptop into an electrical outlet, and you would get your resources that way. How far away do you think we are from that actually being a reality?

Bogomil Balkansky: For this to be sort of the mainstream state of affairs, we're probably realistically five years out at least. With any sort of adoption cycle, it starts with a couple early adopters, and you know some things get the snowballing effect and others don't. To get the snowballing effect, it probably realistically is a five year adoption cycle. The reason I say that, if you look at the adoption cycle of virtualization, realistically, the early adopters like the customer we showed on stage was since 2002. Now it's 2009, virtualization is now squirreling the mainstream, everyone's doing it. It has been seven years between the first early adopter and mass adoption.

Some of our customers are 98% virtual. Even before vSphere. You know, we talked a lot about 100 percent virtualization and vSphere--there's no excuse not to put everything on vSphere. So, if you get to that percentage of virtualization, it means that you have already created this kind of shared compute plan where applications have severed the tentacles of complexity.

So, 98 percent virtual means you've upgraded to the shared compute plans, abstracted the applications. You're in good shape to get very close to the notion of IT as a service. So, what are the additional missing components of that? There are probably two elements. There is some notion of you as IT being able to establish this catalogue of services that you provide, exposing this catalogue of services through some sort of self-service model, where somebody internally can go to this web page and browse for things, so just like if you have to order a wireless service you go to some website, 100 minutes, $30, 1500 minutes, $150 or something like that. So the analogous self-service capability, and then the associated chargeback, kind of the economic sort of provider/supplier model that needs to go with that.

Once you have implemented this internal compute plan, put most applications in a VM, then you really need to, to complete the picture, you need to establish or implement these additional sort of new management technologies. And so I feel like there are customers who could very legitimately claim this soon, as in the next 12-18 months, but for the majority it would take quite a bit longer.

Jeff James: That would be an interesting group of people to talk to: the VMware IT department.

Bogomil Balkansky: Taylor is here. Taylor Stansbury is our CIO--we are 97 percent virtual. And we will be 100 percent in a few weeks, I was told.

Jeff James: Well I think that's it. I appreciate your time.

Bogomil Balkansky: Yep, pleasure.


Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.