A reality check for OpenStack

Last week I caught up with Rackspace president Lew Moorman at the GigaOM Structure event to promote OpenStack, the open source “cloud operating system” started by Rackspace and NASA.

Moorman’s pitch went something like this: In the quest for a more automated and agile data center, we need not slavishly follow the cloud leader, as Eucalyptus has done emulating Amazon Web Services, or get locked into proprietary clouds from VMware or Microsoft. An open source alternative needs to emerge — and it has, in the form of OpenStack.

[ For an early take on OpenStack, check out “OpenStack wants to be your data center OS.” | InfoWorld’s experts take you through what you need to know about the cloud. Download our “Private Cloud Deep Dive,” our “Cloud Security Deep Dive,” our “Cloud Storage Deep Dive,” and our “Cloud Services Deep Dive.” ]

“What I would argue is that there is going to be a Linux in this market,” Moorman told me. InfoWorld explored precisely that issue in “Is OpenStack the new Linux?” by contributing editor Oliver Rist just a few weeks ago.

In that article, Rist was clearly captivated by the phenomenal community excitement around this sprawling open source project. But he also encountered nasty industry politics, along with customers who had evaluated OpenStack and walked away nonplussed because the bits aren’t fully cooked yet.

With the buzz around OpenStack reaching a deafening level — pretty impressive for a project less than two years old — I felt it was time to ask Moorman some key questions about OpenStack’s progress and prospects, given Rackspace’s key role. The following is an edited version of the interview.

Eric Knorr: With OpenStack, you’re biting off a whole lot more than Linux did. You’re talking about storage, compute, networking, authentication, and more. It took Linux maybe 10 years to go from a hobbyist’s thing to an enterprise server OS. How can you expect OpenStack to be usable in the near future if you’re trying to cover so many bases?

Lew Moorman: I think one of the things the [OpenStack] Foundation is going to wrestle with very early on is where does OpenStack end. I am more of the point of view that we should keep it pretty simple in the core elements and then create a framework for extensions that can plug in very easily.

What I hope is that we get very mature in the basics of compute, storage, and networking — and even those are pluggable. Quantum [the networking component] allows you to use multiple different network nodes and plug in multiple different technologies. So it’s built as a modular framework, and that’s at the core of how most people think about it.

No one is building, say, an actual SAN product. We’re talking about a framework into which those can be integrated. OpenStack is not really trying to re-create everything in IT. It’s just going to control the data center, and all the components are going to integrate through this framework.

Knorr: Can you say at this point where the core OpenStack services stop? Essex, the last release, included OpenStack Identity for authentication and OpenStack Dashboard for monitoring and self-provisioning. By default, are those part of the core, too?

Moorman: I think those are critical components, and how that unfolds and what is added is going to be determined by the technical board [of the OpenStack Foundation]. I think that everything you deploy in the cloud world is going to have those components. So is it a lot to take on? Yeah, it is. But it is going to be deployed, in my mind, in every data center over the next 20 years.

It’s hard, yes, but I think everyone said the same thing about Linux back in the day.

Knorr: Even as strong an advocate of OpenStack as Chris Kemp mentioned that in his Nebula appliance, OpenStack is only one of 50 technologies in there.

Moorman: But that’s because it’s a framework. Different storage technologies, different networking technologies — we’re using Nicira, but you don’t have to use Nicira — but the framework is common.

Knorr: What if, rather than things going in a Linux direction, they go in an Android direction where everything gets customized so much by individual vendors offering products based on OpenStack that you lose that common denominator? Don’t you see that as a risk?

Moorman: Could that happen? I think it could. But I think what most folks are actually working on is ease of setup, ease of management, and integrating some of the components where you have to make choices so it’s all simple and out of the box. Making storage choices for you, making networking choices for you, and helping you deploy it with the click of a button.

But the APIs are exposed, the features are there, they’re using the core technologies — I think that is what’s essential. Will every single OpenStack cloud be identical? No, they won’t be. But I think there will be a huge level of interoperability possible and more than there could be across totally different stacks.

Knorr: One thing I’ve heard floating around out there is that OpenStack is really for service providers, such as HP Cloud, rather than for enterprises. Do you agree, or do you feel OpenStack is aimed at enterprises equally?

Moorman: I think it’s going to be used very widely in enterprises. I think what we have now is wide proof-of-concept happening. OpenStack has been in its nascent phase and is about to move into its delivery phase, and I think you’re about to see products and services that make things easier for enterprises. And I think enterprises are extremely interested.

Knorr: It seems like there’s an issue at this point in time of managing expectations. Some of the people we’ve talked to that have experimented with it have said that it’s not fully productized.

Moorman: You’re right. I think the common reaction is “not quite ready for us.” I think this is where you’re going to see a lot of progress over the next few quarters. We can see it’s going to get better and accelerate. I think that’s what has people very excited.

Knorr: So for that to happen, they have to be willing to get the vision and ignore the details over the short term?

Moorman: Look, the truth of the matter is that in any enterprise they’re still figuring out their cloud strategy. It is in the very early days. Right now, anyone who is doing cloud internally is doing it in small, isolated places. People want to make choices for the long term. They see the critical mass around OpenStack, and that makes it a much easier choice.

Knorr: So to extend the Linux analogy, it’s highly unlikely that someone would just download the OpenStack bits and try to get it up and running in their own, right? Is that like going to SourceForge and downloading the Linux kernel?

Moorman: Well, the full OpenStack code is available. There are Chef recipes, and there are other tools that have been built that help you pull things together much more easily. There are definitely companies that are going to make that easier. But there are also going to be free options that make the open source version run pretty easily and pretty well.

Knorr: Last question. Why did you decide to set up a separate OpenStack foundation rather than handing it off to, say, Apache?

Moorman: We’ve been effectively running OpenStack like a foundation for two years. It’s been a technical meritocracy from the beginning. We felt it would be too disruptive to change everything, since we’ve been doing it on our own, and the process has been working well for two years. OpenStack is one of the fastest-growing open source projects in history.

This article, “A reality check for OpenStack,” originally appeared at InfoWorld.com. Read more of Eric Knorr’s Modernizing IT blog, and for the latest business technology news, follow InfoWorld on Twitter.