Sometimes, building software is like playing jazz — it’s all about variations on a standard.
Take, for example, this week’s OpenStack Summit in Vancouver, BC.
At this event, big companies like Red Hat, Canonical, HP, and IBM stand alongside more specialised startups like Mirantis, SwiftStack, and Bluebox to show developers how they have taken OpenStack — a free software project to build cloud computing infrastructure that can trace its roots back to NASA — and used it as a core to build something really cool.
OpenStack is one of the most successful open source projects in history, meaning that thousands of developers from all over the world, representing many top tech companies, have contributed their time, talent, and expertise to building the software over the last five years since it launched.
HP, IBM, and Rackspace all use OpenStack behind the scenes for at least some of their public clouds, where applications run at huge scales on data centres hosted elsewhere. Meanwhile, Red Hat, Canonical, Mirantis, and many others offer their own versions of OpenStack optimised for running in a traditional data center, as well as services to help companies use it.
There are some success stories, too: Huge businesses like Walmart and Comcast came to the summit to talk about how they’re building more efficient data centres using the souped-up, enhanced version of OpenStack that Canonical baked into its popular Ubunu Linux server operating system.
But the thing about OpenStack is that on its own, it’s not complete: Just today, a Gartner analyst described the main OpenStack code as “a science project” and not something to bet a company on, The Register reports.
Business Insider has heard from several sources that the core, so-called “vanilla” OpenStack code requires a lot of work (sometimes, too much work) to get up and running — something acknowledged by the OpenStack community.
“You shouldn’t need a PhD to use OpenStack,” said HP Helion Senior VP of Engineering Mark Interrante on stage during a keynote session at the OpenStack Summit.
All eyes are on OpenStack, maybe now more than ever. Venture capital interest in OpenStack-based startups has cooled off, and once-prominent OpenStack pioneer Nebula, which tried to make it easier to deploy the software in a data center, shut its doors just recently.
Meanwhile, Amazon Web Services, Microsoft, and Google all have booming cloud computing businesses, with Amazon counting $US6 billion in revenue for its own cloud platform in the last year alone.
The success of those three clouds gives the vendors that make up the OpenStack Foundation validation for the principle that there’s a whole market out there. But to take advantage of this big chance, they’re going to have to work with each other to define that standard, even as they scramble to compete for businesses.
The real tough nut to crack here is “interoperability.”
Back when OpenStack was first conceived, one of the big selling points was that all of these different clouds from different vendors would be compatible with each other, which would ensure that you’re never stuck with a bad product.
Every vendor was to be free to build the product that they wanted, for the niches they wanted to sell to. And if a customer didn’t like it, they were free to leave. Where Microsoft and Amazon were difficult and costly to leave, requiring a ton of work to get the same applications running elsewhere, OpenStack would be a common platform that encouraged user freedom.
At least, that was the idea.
In practice, that hasn’t always worked out so well. Vendors were putting lots and lots of proprietary code and features on top of OpenStack in order to get it ready for prime time, which was necessary to get it running, but also meant that none of these OpenStack clouds would work with each other.
This week, that changes for the better: HP, IBM, Mirantis, Red Hat, SUSE, SwiftStack, Ubuntu, and VMware all announced that they have passed a battery of tests designed to ensure that their OpenStack technologies all work with each other, at least partially, reports Fortune.
The announcement doesn’t guarantee total compatibility between these services, but it addresses one of OpenStack’s biggest pains, and at least indicates that it won’t be quite as hard to move between their services as it might otherwise.
The bottom line is that to fulfil the promise of OpenStack, these big vendors are going to have to work together. Amazon, Microsoft, and Google have a tremendous market lead, and a lot of their stuff works well enough for the vast majority of businesses, as proven by their bottom lines.
But they’re also businesses, and they’re competing with each other for customers just as much as they’re making a big show out of cooperating on the development of OpenStack. Sooner rather than later, one of them hopes to hit on the variation on the standard that turns into a huge hit.