Also posted in the OpenStack Silicon Valley blog
“History shows that developers always win,” Mirantis CEO Adrian Ionel told the crowd assembled at OpenStack Silicon Valley this week, “because they create the future. They are the tip of the spear.” OpenStack, he argues, needs to be more developer friendly if it’s going to get past the stage of pilot projects and proofs of concept.
Ionel pointed out that while OpenStack is rising in Google Trends and job listing trends (particularly against Amazon EC2) and revenue has exploded — Mirantis has gone from closing about $1 million in deals every month to closing that much every week — the number of pilot projects doesn’t really give a good indicator of the market. “We need to look at actual customer workloads and developer uptake,” he says, “and there’s much work to be done.”
As an example, he pointed out how developers feel about Digital Ocean, which has very quickly become the #4 web hosting company on the planet. Showing comments on Hacker News, he explained that developers like the hosting company’s interface because it’s dead simple, elegant, and simply works. “Just like OpenStack, right?” he joked.
He then pointed out the fact that Docker has had more than 20 million downloads in four months. “Why? Because it’s insanely easy, with huge benefits for developers.”
“Whoever gets the most developers,” Ionel says, “wins.”
When it comes to OpenStack, he says, the question is, is it for people who are more interested in control? He called these the “DIY dead-enders”.
While many arguments in OpenStack developer circles revolve around the choice of monitorning software or what hypervisor is available, Ionel pointed out that developers don’t care about those things. They don’t care about whether they’re a choice of storage systems on the back end, they care about whether they can store their files. They don’t care about how the network scales, they care about whether it works.
Most of all, Ionel pointed out, developers care whether there’s a reliable, quality, easy-to-use API they can write to. They also care about feature velocity (but only the features that they care about) and the ability to move their applications around without having to rewrite them. “That’s why Docker has been so successful.”
Ionel feels the OpenStack community needs to take steps in this direction to be successful:
Focus on the APIs, since that’s how developers interact with OpenStack. They must be clean, and they must be consistent.
Invest in ease of use and even more flexible plumbing so that other systems can plug into it.
Don’t move up the stack; partner with other companies instead. AWS continually moves up the stack, he points out, adding features that cannibalize other markets, such as Load Balancing as a Service. Rather than “following the AWS playbook” Ionel suggests partnering with complementary services.
Enable workload mobility to other platforms.
Reshape upstreaming to foster open competition.
This last idea is perhaps the one with the biggest implications for the OpenStack community. It means, he explains, enabling alternative solutions within teams, and letting the market choose. He points out that the “accepted” deployment technology is TripleO, but that only HP, which created it, is pretty much the only one using it. Everyone else, he says, uses Chef, Puppet, and so on.
Any central planning should be advisory, he emphasizes. “We all know how well central planning works in creating business opportunity. Let’s not repeat that in OpenStack.”