Shuttleworth: OpenStack made mistakes, but Kubernetes not a replacement

Shuttleworth: OpenStack made mistakes, but Kubernetes not a replacement

Pity OpenStack. Celebrating its 10th birthday this coming October, the open-source cloud was conceived as the future: the open-alternative to Jeff Bezos’ proprietary behemoth.

A decade on and despite success among telcos and internal IT departments, OpenStack is now being mentioned in the same breath as Kubernetes – just not in a good way.

There’s a burgeoning discussion as to whether it’s possible to wire a serverless cloud using open-source containers instead of OpenStack.

Ten years on and we haven’t moved on, it seems, when discussing OpenStack’s faults and failings – to wit: it’s complicated. There’s a belief you can achieve the desired outcome of a private cloud using the Lego-like Kubernetes without having to make the bricks first.

The Cloud Native Computing Foundation (CNCF) last year reportedly told DockerCon the project’s stewards are looking at a way of using Kubernetes to replace OpenStack – and VMware. CNCF this month accepted a Kubernetes container runtime as an incubation project – Container Runtime Interface Orchestrator (CRI-O).

Is OpenStack being shown the thin end of the wedge by the kind code Avant Guard  whose choices get to determine the platform your IT team is saddled with for the next 25 years?

Ubuntu founder and chief executive Mark Shuttleworth thinks it’s time to take a step back from this counter revolutionary mindset. Shuttleworth, whose Linux spinner was an OpenStack founder and that today builds OpenStack private clouds or those leaving VMware,  reckons there is no “either-or.” It’s a case of the right technology for the right use case.

“I believe OpenStack is important,” he told this month’s  Open Infra Day UK 2019 in London. “It’s become trendy to say: ‘I’m skipping OpenStack and going straight to Kubernetes.’ It’s like skipping salad and going straight to steam – they both solve different problems.”

Speaking after his conference keynote, Shuttleworth levelled with DevClass: “For a crowd of people who hop from shiny to shiny, OpenStack is no longer shiny and other things are shiny. But at the end of the day, what matters is the problem that’s really being solved and I think OpenStack solves some important problems for very large organisations.”

Shuttleworth buys into OpenSack giving you multi-tenanted, private cloud with a software defined matrix for managing large-scale compute with lots of different users.

“Kubernetes is super-useful for modern-designed applications,” he told DevClass.

“So the idea is of there being tension between them, the idea of there being winners and losers is kind of crazy in a world where both play an important role.”

The risk in skipping straight to Kubernetes or trying to build a private cloud simply with Kubernetes and without the elastic framework, storage and management fabric of OpenStack is you end you get stuck with a custom infrastructure that demands on-going support.

“One of the little lies we are telling ourselves these days is you are just doing this quickly,” Shuttleworth he’d told Open Infra Day, “but we are going to live these things. People love to tell themselves: ‘I will do this thing quickly and throw it away.’ But these things are going to be with us in 10-years’ time. Make sure you can come back and maintain these things and security patch them.”

What’s difficult, or has proved difficult for OpenStack is the persistent perception and reality of “it’s complicated.” AWS is complicated, too, but Amazon makes a better fist of hiding that. You’d can Kubernetes with AWS, but you wouldn’t try to use Kubernetes instead of AWS.

“The reality is OpenStack did make mistakes,” Shuttleworth says, looking back over the last decade. “OpenStack is complicated, and for a lot of people that makes it something they won’t achieve so it’s something that’s easy to dismiss.

“Lots of organisations were disappointed by OpenStack because of their first run at it – of throwing a pile of Python at their IT department didn’t work out. If you do it right,  OpenStack gives you a cheaper way to do what VMware does.”

At the end of the day, according to Shuttleworth, you should recognise there’s a risk factor attached to OpenStack but manage that risk as you would manage risk elsewhere in IT.