Sept. 19, 2012 will go down as a big day for OpenStack.
The open source cloud computing platform project began about two years ago as a collaborative effort between Rackspace and NASA, along with a couple dozen other companies and a few dozen developers. Since then, it's grown to include 180 official partnering companies - including some of the biggest names in technology, such as AT&T, Cisco, HP, Dell, IBM, Red Hat and VMware - and an ecosystem of 400 contributing developers and more than 5,600 users.
A CLOSER LOOK: Who makes up OpenStack?
IN OTHER NEWS TODAY: 10 ways to celebrate Talk Like a Pirate Day
One company seems to have stood out above the rest when it comes to OpenStack. Rackspace is not just a founding member of the project, but its president has been an outspoken advocate of the project and the company has made significant investments in powering its own cloud offering from the OpenStack code. In some ways, it has felt as though OpenStack has been a Rackspace project.
Today, that changes, at least technically. Rackspace has legally ceded control of the project to a newly formed OpenStack Foundation, which is now responsible for the OpenStack trademark, policy, technical development and project advancement. In the late summer the OpenStack community held an election to nominate and select a 24-member board of directors, which will govern the foundation, along with a a chairman, vice chair and executive director.
So what's next? Network World sat down with the new leaders of the OpenStack project, Chairman Alan Clark, director of industry initiatives, emerging standards and open source at SUSE, and Lew Tucker (pictured left), vice president and CTO of cloud computing at Cisco, who serves as vice chair.
What goals do you have for the newly created board of directors?
Clark: I would categorize what we're doing in two steps. The first is establishing the basics and covering the bases. We've spent some time getting an executive director in place (Jonathan Bryce, formerly of Rackspace), and we're still in the process of getting all the fiduciary models, assets and trademarks rolled over. That stuff has taken a lot of time up front but with the Foundation launching, that's a big step in the right direction. Next, is working on the priorities of the foundation over the next year, which involves strengthening our ecosystem, accelerating OpenStack adoption and making sure we're delivering the best cloud software out there.
Tucker: OpenStack is gaining huge momentum right now, which is witnessed by the 180 companies and 5,600 members. So the foundation's role and responsibility is to provide the guidance for that and make sure we can really focus on how OpenStack gets used and plays out in the marketplace. We're really driving the adoption of OpenStack more than anything else. The technical committees and leads are responsible for the development of the software and the platform itself.
Who is OpenStack for? Some have said this is a play for service providers to create their own cloud offerings. Is this code meant to be taken up by an enterprise and used directly themselves? Is it optimized for any one group?
Tucker: We've got companies in both of those categories today. We really think about this as a platform and if you think about the platforms that have really made it in the world - Linux being one of the foremost , Java being another -- it's a horizontal play and you really want it to be adapted for the different kinds of use cases. We're really focusing on a broad range of use cases rather than a particular niche in the cloud computing world. And that gets back to the ecosystem because a lot of the power of OpenStack will be the other products and services that will be built on top of OpenStack. That will allow it to go into different verticals and allow it to be applied to different situations. Meanwhile, we're hoping the core platform and the core projects are the foundational pieces that will underlay all of this.
Clark: It's really not optimized for anyone, but instead one of the reasons people participate is to make sure that core product operates sufficiently in their environment. That's why we have so many companies contributing. If you have an interest in one particular hypervisor, for example, then companies with that interest will make sure the code works in that environment. Someone else has an interest in a different hypervisor, so they contribute that. When that happens, everyone wins.
Some enterprises that may be interested in OpenStack may be wondering when it's right for them to jump in. On the one hand, they want the code to be developed to a point of stability where they know it will not be significantly altered if they make large investments in it. On the other hand they want it to be mature enough for enterprise use. So where is the code now and how do you see it developing in the near and mid-term future?
Clark: The code is stable now and a reference to that are companies like Rackspace and others that are running their businesses on this code now. The core is there and it's very well proven. Where's it going? What barriers need to be reduced? That's why one of our priorities is to accelerate adoption and get more input on those questions. It will come down to the different companies and users involved in the project having different perceptions of what those barriers are and which areas need to be improved, but I think there are some commonalities around improving ease of use, ease of deployment and interoperability.
Tucker: One thing that's really interesting to look at is on OpenStack.org there are blueprints that are filed by various companies indicating the projects they would like to see developed. In many ways you're getting an early read on where the industry as a whole is going by looking at these. There are a lot of projects being put up about making OpenStack easier to use, making it more manageable at scale and then expanding the capabilities to cover a wider range of cloud workloads.
The code is still developing though. The next release of the OpenStack code, dubbed Folsom, which is set for release at the end of September, for example, will have what some consider a major advancement by bringing Quantum, the virtual networking project, into being a core project within OpenStack. What's the significance of that move?
Tucker: Quantum we felt was sort of a missing element in many of the cloud stacks out there today that focus on compute and cloud storage. We really felt a complete cloud platform would have compute, networking and storage services. It seems obvious in hindsight, but bringing in a network service does a couple of important things. First of all, networking is undergoing a big transition in the marketplace as we're seeing things such as OpenFlow and software defined networking come into play. By essentially refactoring compute so that networking is now a service, that will make it a lot easier to accommodate a lot of the changes happening in networking. This is an area that we think will undergo some rapid changes in the coming years and Quantum allows us to create network capabilities as applications, such as multi-tier networking, while allowing us to insulate that from different vendors and the underlying infrastructure.
Is there a connection between OpenFlow and OpenStack?
Tucker: Sure, absolutely. Some of the implementations that we're seeing are OpenFlow controllers being used as a plugin to Quantum. You can think of Quantum as a hardware abstraction layer model where you can plug in different technologies. Some of those, for example, are the variety of OpenFlow controllers coming out. This allows you to use OpenFlow technology in OpenStack. Cisco has a plugin, Nicira has a plugin and several others are coming out.
In the past year we've seen the open source cloud landscape develop in some interesting ways with projects such as Apache CloudStack, which is backed by Citrix, and Eucalyptus gaining prominence. One point of differentiation those two projects seem to make is around interoperability with Amazon Web Services APIs. Do you think there should be more focus on integrating OpenStack with Amazon?
Tucker: This is an active discussion in the technical committees, but we have interoperability with Amazon EC2 (elastic compute cloud) APIs. The real question going forward is as we go further in cloud computing there are some critical innovations that I think need to happen. We would like to maintain that compatibility and interoperability with Amazon by in large, but we also find that we need to have an outlet with which to innovate. So we're actively pursuing this sort of dual strategy where we have this OpenStack API and we also have a compatibility layer that allows people to migrate workloads and make their systems interoperable with Amazon.
One somewhat controversial topic around the nascent OpenStack board has been the recent vote bringing VMware into the foundation. Some have said that OpenStack is viewed as an alternative to VMware they're worried that with the company being part of the project it will muddy the market perception around OpenStack as an alternative to VMware. Are you worried about that at all?
Clark: I'm not worried about that at all. I know there have been other people involved in the project longer and they may feel differently, but experience shows that as a project has momentum and is gaining momentum, people will join in and participate. Their reason for doing so will vary because their business plans vary, but they do contribute. A prime example of this is with Linux. Microsoft stood up for years and lambasted Linux, but go look at the amount of contributions that Microsoft made to Linux last year, they were in the top 10 in contributors to Linux. So will VMware compete? Yes. Will they contribute? Yes. That's just part of the ecosystem.
And both can be done?
Clark: Yeah, we're all competitors one way or another. But while we compete we also collaborate and find common ground. It's already happened and it will happen even more with VMware being involved. They'll compete on different layers, they'll contribute on common layers and the result is the customers and the community will benefit from that.
Some have said OpenStack is the Linux of the cloud. Is that fair, is that the end game goal?
Clark: The momentum of the project is huge and it's clearly pulling everyone in, so I see a lot of similarities in that regard to Linux and I think this very much will become a standard for cloud computing in the open source world. The end game is a long ways off; we're still very early in this process.
Overall, how would you assess where the project is so far? Over the past two years it's built up an ecosystem, but still only a handful of service providers and users have launched major OpenStack-powered clouds, with Rackspace and HP being the notable examples. Do you wish this had been developing faster? Are you happy with the progress to date?
Clark: I'm extremely happy, I mean if you think about what's happened in the last two years, we're now on the sixth release of the software this month, and those releases have been on time. The community has been growing by leaps and bounds. You can look at a specific area about how many hosting services are using it, but you can also look at the ecosystem. All the major Linux distributors are on board and releasing their own versions, so to me that looks like pretty good progress.
Tucker: I'm very pleased as well. Of course as an engineering manager, I always want it to go further, go faster. We have a lot of work yet to do in front of us making it easier to use, easier to install. And I think as we start to see more of the distributions come out, that's a lot of what they'll be doing - making it easier for customers to consume. And I think that will also be able to bring it into a lot of new areas. I think there will be as many deployments of OpenStack used outside of the public cloud space as there will be implementations used to host storage services or other media services. There's no faster development and deployment platform right now than the cloud. A lot of the newly developed applications today, as Amazon is showing, are being done in the cloud. With mobile services coming out and being a big play, I think they'll find a real home on OpenStack.
Clark: I completely agree and the only thing I'd add is the fastest and best way to get there is through open source development.