The OpenStack community is of two minds. At first glance, it is extremely popular, as is evidenced by the recent decisions at Citrix, HP and Red Hat to support it. But upon further inspection, it becomes clear that this open-source community and project are still in their infancy.
Dave Bartoletti, senior analyst for infrastructure and operations professionals at Forrester Research, agreed that OpenStack lacks maturity. Specifically, he said, the interfaces between the various components that make up OpenStack are still too fluid in terms of their code. Most internal OpenStack APIs, such as the Nova API for allocating compute, cannot even tell you what version they are via an API call.
Bartoletti was at the OpenStack Conference last month, and he said the event showed a vibrant but still-young project. “At the conference last month, what was nice is they brought some actual customers in to be heard. There were plenty hoodies and suits at the conference, too. There were plenty of suits getting involved, with HP and Red Hat throwing their weight behind the project.”
That sentiment is echoed by Red Hat itself. Greg Kleiman, senior director of strategy at the Red Hat Storage Business Unit, said there is no denying his company’s commitment to the project overall. “OpenStack is absolutely the Red Hat strategy for cloud,” he said. “We are 110% in on OpenStack in terms of our strategy to the cloud. Given that we’re bringing all our assets to bear across our portfolio, we think customers are going there, and therefore we want to be there.”
To that end, Red Hat has committed to helping build out OpenStack Storage components Cinder, Glance and Swift, and a new project proposed by NetApp for storing raw files in OpenStack Storage. With those three storage mechanisms in place, developers and admins will be able to choose between storing blocks, files or objects. Additionally, Red Hat plans to help contribute to a project that will allow raw instance storage within OpenStack.
Beyond that, Red Hat is looking at OpenStack as a way to expand its middleware offerings, as well. “Storage is a very big part of OpenStack and cloud,” said Kleiman. “We think Compute is another big part of OpenStack and cloud. We think there’s clearly a bunch of middleware pieces that need to be there both in a single instance and at a high level, and that’s where we’re bringing all our JBoss experience to bear.”
And while Red Hat and others are touting early-adoption customers, such as Best Buy and Bloomberg, Bartoletti said that there is a danger for early adopters in the form of removing themselves from the open-source pool due to the necessity of customization. Building on OpenStack this early, he said, could result in forking yourself away from the project.
“They’re doing it because it’s exciting. It’s a vibrant, active community,” said Bartoletti “The Comcast guys said the reason they wanted to [build with OpenStack] was they liked the community aspect of it. To them, they liked that over a more packaged vendor solution where they’d be limited in what they could do.”
Ultimately, while OpenStack still has some distance to cover before it could be considered mature enough for enterprise use out of the box, it’s still offering a very compelling use case to businesses as a way to bring shadow IT under control.
“What was really interesting,” said Bartoletti of the OpenStack Conference, “was almost all the big customers that spoke discussed the shadow IT department. ‘We’ve got developers using Amazon. They’re using it because it’s cheap and easy. We’d like them to develop internally. How can we offer them something like Amazon?’ ” asked many managers at the show.
“What Best Buy was focused on,” said Bartoletti, “was that they have a bunch of teams building in Amazon. Then, they bring it back inside and the internal team has to rebuild these environments and match them. It’s more a way of saying, ‘Let’s build our own development platform that looks Amazon-y, has a self-service portal and some interesting blueprints and templates, and try to offer developers the same speed internally while leveraging the same infrastructure. I don’t think that’s for everybody. These guys were definitely early adopters.”