Red Hat's Jim Whitehurst: The Challenges of Competing with Free
Red Hat's Jim Whitehurst: The Challenges of Competing with Free
Red Hat, which bills itself as the world's leading open-source solutions provider, has managed to make free software pay by gathering, extending and packaging Linux and complementary open-source components into certified and supported products that are ready for enterprise consumption.
As the focus of IT attention shifts to new platforms, including virtualized environments, Red Hat has turned to an IT industry newcomer, former Delta Airlines Chief Operating Officer Jim Whitehurst, to guide the open-source leader.
eWEEK Labs Executive Editor Jason Brooks recently spoke with Whitehurst, now president and CEO of Red Hat, to discuss the company's expansion to new platforms and markets, and the challenges around building solid business models for products that can be obtained for free.
It seems like in the last couple of months there's been a real uptick in the announcements and releases that have been coming out of Red Hat.
Yes, actually, we'll deliver more new products this year than in any year in our history. It's really a lot around our cloud and grid computing strategy. There's management, there's security, there's virtualization-and all the tools that go around that. These are all key components as we think about grid and application mobility. So there is a lot going out, but it's part of an orchestrated deployment.
It seems that the main focus of Red Hat's cloud strategy is helping enterprises build a cloud infrastructure of their own. Is that the case?
We play a key role between hardware and applications-be that at the operating system level or the application server level-so we are very focused on starting out with the end in mind. If you're a customer, you should be able to run your application however you want to-be that bare metal, on a hypervisor, on the cloud or as a virtual appliance-and we should be able to allow any of those to happen.
What we basically have done, and we guarantee: If an application is certified to run on RHEL [Red Hat Enterprise Linux], it's certified to run bare metal, it's certified to run on a hypervisor, and it's certified to run, right now, on the Amazon EC2 cloud.
For us, one of the key things-given that we come from a heritage of working in the data center with large enterprises-is making sure whatever deployment model [companies] decide for their applications will work.
Now, a lot of that is work around the cloud, because, obviously, we already have bare metal and hypervisor. But, [the cloud] really does look like a key deployment option that our customers want and that in general looks like a direction the industry seems to be moving in, so we need to be there for our customers and for our ISV partners.
Expanding Red Hats Base
Do you see the movement toward the cloud as an opportunity to start moving to smaller customers, expanding your base?
Certainly. As we move further down market, we clearly don't have as large an ecosystem for small business as does Microsoft. So one thing that certainly accrues to our benefit is fundamentally changing the computing paradigm. Consuming computing cycles via a grid in the end works for us because grids are highly likely to run Linux. We need to make sure it's our Linux, but it will almost for sure run Linux.
So, certainly, by changing the computing paradigm, it gives us an opportunity to significantly expand the company coverage we can have. That's not to say we don't sell now to small business-we sell to a lot of Web small businesses, LAMP [Linux, Apache, MySQL, Perl/PHP] stack, etc.-but I don't think my dentist's office uses Red Hat Enterprise Linux. However, in a grid-based model, there's a good chance they would. And so it does allow us to significantly expand our base.
I think it may also be our heritage, but we don't get religious in terms of how exactly the cloud develops. With one of our very large customers that runs over 10,000 servers, we've done tests live migrating stuff to the EC2 cloud and back, and, you know, all that works great.
But the CIO's point back to us was [that we were] getting into nickels in terms of the difference for him to source at that kind of scale versus Amazon or anybody else to source, and so he says, "I don't want to pay the average cost. What I want to be able to do is [have] you find me one of your other customers-like a bank that's not using their systems as much at night or a disaster recovery site-and I want to be able to migrate over and use their cycles. I'll pay them, not average costs, I'll pay them the marginal cost. They're not using the cycles at all, they'll get paid, and I'm actually able to source cheaper than Google or anybody else who has fixed costs."
And so, we're working with customers-this is still kind of early on-in thinking about how we not just make grids into public utilities, but also how we potentially open up customers to use each others' infrastructure.
We don't know what the model is; we just want to make sure it's doable. We want to make sure we have the technology in place to secure it. If you're a bank, and you're going to let somebody else come in at night ... you want to make sure you have airtight, bulletproof security, right?
You've got to be able to manage all those workloads, all the components associated with that, so we're really working on the technical side now. We think that we, as the leader of open source and because we're open source, can be a trusted partner in doing that. We can't come back and gouge you later, because you can stop paying us.
I understand why people like Google are basically running their own applications, on their grid, and not allowing third parties necessarily to come in. Because, how do you know they work? How do you know they're certified?
Well, we've done that; we've done that heavy lifting. If it's certified on RHEL, it's certified to run in any instance, including cloud.
What about software appliances? Could we be moving to a situation where a customer isn't necessarily worrying about the operating system and the application working together, since the application vendor has worried about that?
That's part of what we're doing-for an ISV, we're making it simple. It's not saying, "Well, I've got to develop my certification and everything else for an appliance here, and then I need to make sure it works on Red Hat, and I need to make sure it works here and here." Do it once, and it's all there. I think appliances will continue to grow in importance. To have all that be one standard certification is powerful.
In the face of free competitors, or lower-cost competitors, will Red Hat need to move toward something like cloud hosting to stay in the middle of the transaction?
I certainly understand that CentOS and other free Linuxes are out there, but as the use cases become more complex-cloud, integration of all the components and virtualization, live migration, security tools, etc.-as integration for anything gets more complex, the more likely you want the people who wrote it supporting it and the people who certify it supporting it.
It's not just about RHEL; it's RHEL, it's virtualization; it's IPA [Identity, Policy and Audit], our security suite; it's Red Hat Network and oVirt, the project around the automation tools around managing the virtual instances.
You start putting all that together and you start saying, "Well, how am I going to pull out one layer and put, you know, a community version in here?" It's a lot harder. So, as the use case around it all becomes more complex, it's more likely that people will see the value in the subscription and the support and the service. And so certainly I think that accrues to our benefit, as well, regardless of the specific business model.
The relationship between Red Hat and CentOS is unique to the open-source world. I happen to think it's good for Red Hat that CentOS exists, but what's your take on it?
I think you're right. I mean, it's very easy in any business to get complacent, and CentOS keeps us on our toes with every single day [because] if we're not adding value that day, the customer can stop paying us. CentOS is there.
Now, again, [CentOS is] not Red Hat Linux-you don't get the certifications, you don't get the updates. It's different-it's a different set of ones and zeros, it's different bits, it's not certified.
That said, [CentOS] has functionality that's similar, and that keeps us on our toes. If we're not delivering value to the customer, they can turn us off, they can go to CentOS, they can not pay us. So, it does keep us on our toes, but I think it keeps us on our toes in a good way.
On a somewhat related topic, I'm curious to hear what your impressions are of Ubuntu.
I'm very pleased that Ubuntu is out there and [increasing] awareness of Linux. I think that's a great thing. We really don't compete with them in any material way. I think there are some very clear distinctions between Fedora and Ubuntu that I think are material and important in terms of our view and the importance of open drivers and open standards.
Ubuntu a Concern?
Microsoft has drawn a lot of strength from its success in pursuing the goal of ubiquity and of Windows on every desktop. At one time, Red Hat was probably what most people would have turned to when getting started with Linux. Do you think that distributions like Ubuntu might be taking over that position, and is that a concern for you moving forward?
There may be a little issue there, but Red Hat Fedora is a massive large free distribution that is out there with leading-edge technology, so I feel comfortable with that. I think that the overall benefit of more Linux out there is certainly better than the risk around [competition from Ubuntu].
You know, we came out and said we weren't going to have a consumer desktop, and some people were thinking, "What are you doing?" And, you know, I guess that a better way to have said that is, "We are not going to have a business model built around a consumer desktop."
Some people will say, "Oh, there are so many desktops out there ... blah, blah, blah." And I look at it and say, "You know, great, but I want Linux on them, and I don't see why anyone should pay for it. Why should you at home pay for a desktop? It should be free."
But we've built, arguably, the only successful open-source business model of any scale because we address the customer's need. We take open-source software, the power of iterative innovation, and make it consumable to the enterprise-the testing, the tuning, the certification, the service-level agreements. If you're on the New York Stock Exchange, those things matter-they matter a lot.
But the average home user? Fedora's never crashed. If it did happen to crash one time, would it be catastrophic? Everyone is used to Windows crashing all the time, and so why should you have to pay for all the service levels and the support for the absolute certification and the certainty that it's always going to run?
If you don't need that, you shouldn't be paying for the desktop. So, again, I draw a clear distinction between where open source should be and where Linux should be and where open-source business models should be. Open source will continue to pervade a lot of areas, but I think where those are monetized won't necessarily directly overlap.
I mean, [Canonical founder Mark Shuttleworth] can obviously do it [charge for Ubuntu]-it's his company-but I don't see the business model relatively.
Again, I applaud Ubuntu; I'm glad they're out there getting Linux out there. I do think they play a little bit closer to the edge, especially around proprietary codecs and those things. ...
My concern there is, and this has died down a lot, [that] Microsoft and others early on tried to portray open source as a bunch of pirates who had no respect for intellectual property.
I would actually argue that people in open source have extraordinary appreciation for intellectual property and the power of intellectual property. Look at the GPL [General Public License]. It uses intellectual property and the protections around that in a very powerful way to keep it open. So true open-source believers understand and value the importance of intellectual property.
I think that even more ingrained in the users of Linux and open-source software than a respect for intellectual property or an understanding of the importance of intellectual property is a desire to hack around problems.
I agree. But, I think as leaders in open source, it's up to us to lead that battle, not to say, "You know, let's make it easier." I mean, we need to make sure we champion that. I think, especially Red Hat, as the established, clear leader in open source, we have a special duty to make sure that we are living by the principles and leading by example. From the patent settlement that we recently did to supporting ODF [OpenDocument Format]-all those things.
That patent settlement was a really good example of how Red Hat is different from, say, Novell, in the way that it is approaching these issues.
Novells Enterprise Desktop
Speaking of Novell, I know that while Red Hat isn't focused on a consumer desktop, an enterprise desktop has been part of your plans. What do you think of Novell's enterprise desktop?
They have made a broader investment in [the enterprise desktop]. I'll be frank: I think we in Red Hat for a long time kind of muddled desktop with consumer, and we really underinvested in the desktop versus what we should [have done]. So, we're redoubling our investments in the desktop because the desktop's important.
Our customers are asking for it, we need to have an excellent desktop, and I think we have a very competitive desktop platform out there. But we have not invested as much or as heavily as others.
If you look back at the core, GNOME, if you look at the contributions, we are still a major, major contributor around the desktop and the components that go into the desktop, and I think we actually have a good one.
In Eastern Europe, you saw the bundle we've done with IBM and Lotus Notes. It's been very, very successful, so we are making real strides and a lot of progress there. It's been a more recent push than I think historically we've had because I think there was this, "Oh desktop-that means consumer, and we're not doing that." Let's be clear: The desktop's very, very important to what we're doing. I use it all the time; it's quite good.
Do you use Fedora or RHEL on your desktop?
My desktop at work, I use RHEL, that's our corporate standard build. At home, I use Fedora.
I think that the things that Red Hat's been working on in the desktop space-and even things not specific to the desktop space, like SELinux-I think the pieces are there to make a very compelling desktop, not just for enterprises but for a lot of consumers, too. People who would be willing to pay for something that's built around SELinux and Stateless Linux that is locked down and that people can feel safe using. This is a big hole for Microsoft right now.
Actually, I laugh as you say that. Have you ever run Fedora off like a USB key?
Yes, I have.
And you know we now have it so you can have persistence on it. There's a huge fight around that because persistence is moving away from this idea of stateless, and the original idea around the key and some of that was let's get people used to stateless. ... We still talk quite a bit about it, and we're marching forward, so we have some interesting things coming up going forward on that.
But the cloud is really exciting because that's the place where it can bleed over. It doesn't matter so much the exact boundaries, or it needn't matter-it's accessible.
Well, it's interesting. Once you do that, then you start saying, "Well, the desktop becomes relatively less important." Right, because your functionality is all off somewhere and it's really your front end. We're doing a lot of work on GNOME and a lot of the front-end stuff. Fedora's a great desktop. I mean, the reason I have [Fedora on a USB drive] in my pocket is I often, instead of using the corporate standard build, will boot with this because I'm a big Fedora user-[I have been] for years.
IT's gonna come after you.