Red Hat CEO talks virtualisation, Linux and the cloud

Summary:Red Hat chief executive Jim Whitehurst discusses why he thinks rival VMware will fail, how the financial crisis will be good for open source, and why cloud computing will be the future

At a relatively sprightly 40, Jim Whitehurst joined Red Hat from US company Delta Airlines, officially taking the reins on 1 January of this year.

The Red Hat chief executive has previously named cloud computing as his number-one priority. When caught up with him recently, he added virtualisation to the top of that list.

Q: When you were last interviewed by CNET ['s sister site], you said your central priority was cloud computing. Why do you see this as important?
A: I do believe we are at an inflection point in the IT architecture. I am not sure it is quite as big as going from mainframe and terminal to client/server, but it's certainly a significant change in architecture.

Basically, if you look at the economics, there is no question that it is cheaper to generate a computing cycle in a centralised farm of commodity server hardware than it is in the current client server environment. That includes utilisation, the cost of hardware, the maintenance and support, and so on. I'm a big believer that economics will win in the long run, as we see more and more functionality move to large server farms that are centrally managed.

I guess whether they are internal private grids or clouds, or whether those are public, or semi-private, that's still to be determined, because that's a lot around specific business models. However, there is no question that functionality will drive a move to more cloud-type environments. I think building a new architectural cloud around propriety software is inadvisable at this point in time. If you're building a new architectural paradigm, why would you do it on propriety hardware and risk lock-in?

You offered virtualisation in RHEL [Red Hat Enterprise Linux] 5, but you aren't known for it. How do you plan to compete with VMware, Citrix and now Microsoft in this space?
VMware, no question, was first to the market. I frankly have run into no customer yet who is running XenSource.

We actually have a lot of traction. We are gaining quite a bit of momentum with our virtualisation. This is one place where Microsoft and Red Hat agree. Virtualisation should be part of the operating system, not a separate layer. I say that for several reasons.

First off, a big, big chunk of virtualisation is hardware enablement, and hardware enablement is something already being done by us — and the Linux community. The Linux community spends a lot of time — and hardware manufacturers and chip OEM manufacturers spend a lot of time — writing drivers and enabling hardware already in Linux and in Windows.

So adding a whole [additional] layer? To enable hardware and be involved in those technology roadmaps just doesn't make a whole lot of sense.

Secondly, a lot of long, hard work has gone into a whole bunch of areas within the operating system that are inherited by virtual guests. Let me take a simple example, security. The security regiment that is in [Security Enhanced] Linux was originally heavily contributed by the NSA. It is certainly considered the most secure operating system. The Russian military has RHEL certified as the most secure operating system available, period.

A virtual guest coming out of an RHEL [instance] inherits those [security] characteristics. The idea of trying to go back and completely rewrite a security paradigm to work with a hypervisor — I guess that's possible, but it'll take five or 10 years and thousands, or tens of thousands, of man hours. […] The Linux kernel is 9,000,000 lines of code. The average hypervisor is 35,000 lines. Our belief, and I think Microsoft agrees, is that you have all the work and engineering that has taken over a decade or more to put into the operating system. To now decompose it all, throw away the operating system and recreate those in all kinds of layers, is crazy.

It's a bit like, 10 or 15 years ago, people used to buy a separate TCP/IP stack, but these days you don't buy that, you expect that to be in the operating system.

I think in the next two or three years you will hear people stop talking about hypervisors, because they will just be ubiquitous in operating systems. I think Microsoft and us are really…

Topics: Tech & Work

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

The best of ZDNet, delivered

You have been successfully signed up. To sign up for more newsletters or to manage your account, visit the Newsletter Subscription Center.
Subscription failed.