A consistent theme of this week’s OpenStack Summit in Paris has been a tight focus on business benefits and vision, which has set the conference apart from many developer-centric events.
At times it has felt more like an evangelical US C-suite jamboree than an inaugural summit in Europe for the global cloud developer community.
With cloud prices under pressure again, thanks to Google's latest price cuts, the open source community is well placed to capitalize on a more business-focused, lean and non-proprietary approach to rapid cloud deployments.
Their cause has been helped by blue-chip users with memorable stories to tell: 'super user' CERN deploying OpenStack to find the missing 95 per cent of the universe, BMW developing its i8 supercar with the aid of an OpenStack-based datacentre, and so on.
OpenStack Paris 2014 feels like an event with a well-planned and executed message – the community that thinks like a corporation.
That message is about getting developers to talk the language of business, about moving IT from backroom to boardroom and asking a generation of coders to think about end-user benefits and real-world applications.
So does this business focus represent a policy decision by the OpenStack Foundation?
Foundation COO Mark Collier says:
My perspective from talking to a number of developers who have been working with OpenStack for years is that they all now work for companies that are now making money from OpenStack, and you run into them and it's like, 'Oh, you're wearing a suit? What's going on?' And the response is, 'I've been talking to customers'.
So the language they use is now very different. Back in the early days of R&D it was 'What can we build? Let's automate some servers, that sounds cool', but now they're hearing straight from people who are getting real business value and they're starting to use the same language themselves.
It's a natural evolution from the 'science project phase' to the real world, where they can see how big an impact they're having.
This increased maturity and shift in perspective is inevitable, he suggests:
We're on our tenth release now [the Spring/Fall upgrade cycle means two releases a year], so there's a maturity level that means the business and the technical sides of it aren't as separate as they once were.
Customers tell us that their business moves faster now, and developers think that's really cool. They hear what their work is doing in real life and that gets them excited.
Executive Director Jonathan Bryce adds:
I think all of our speakers said 'thank you for building this and contributing to it'. But they have a wish list too, so it's more 'Thank you, but now can you help me?'
Developers do pick up on all this more than they're given credit for. This kind of software is not a science experiment or a college assignment, and so when you find out how it's being used and the ways in which users would enjoy using it even more, then I think it connects all the dots for both sides.
Not all positives yet
But as previously reported by diginomica, not all of the messages coming out of the summit have been so positive, with some high-profile users and contributors complaining of the constant, packaged-software-style upgrade cycle, and questioning whether that is necessary in a fast-maturing space where build-out and traction are both well established.Bryce contends:
The criticism is not that surprising. With people running OpenStack in production environments they want to have some stability. The criticism has been trickling in for the last couple of releases and the technical community takes that feedback. I guarantee that's something that's going to be talked about at the design summit.
So is the answer changing the release cycle? Maybe. There isn't an answer to that today, but it's going to be talked about this week at the design conference. But on the upstream side, the community side, we don't want to disadvantage one group of users over another. We're providing options for everyone getting their needs met.
There are companies like Time Warner that like the rapid release cycle because what they're doing is advanced on the networking side and they need the most advanced features. They're already running components from [next iteration] Kilo, so they're ahead of even the most recent release.
But there has been a definite shift from feature-adds to making the technology easier to operate. [Latest release] Icehouse offered a major change in how upgrades are managed, and Juno does too. So now we're at the point where you can do live upgrades without disturbing the workloads, and a stable branch management system was introduced last year.
To put this in context, according to OpenStack, 67% of OpenStack users run one of the two most recent releases – a percentage that Microsoft would kill for – and 47% of usage is in live production deployments. And bear in mind: this is all software released in the past 12 months.