OpenStack is in a state of flux. The open source cloud infrastructure has always been dogged by issues regarding its value, complexity, cost, and a lack of skills. But now it faces a new challenge within its community that could herald a make or break period for the technology.
The cloud infrastructure has to some extent overcome many of the challenges it has faced, and dispelled many a myth in the process, but moves from some of the largest vendors in the ecosystem have threatened to de-rail the project.
HPE and Cisco are two of the most notable companies to have shifted their position with OpenStack, whilst Intel has also withdrawn funding in the past year.
Mark Shuttleworth, founder of Canonical and soon to return to the position of CEO, spoke to CBR recently about the health of OpenStack.
On HPE’s position Shuttleworth said: “Here’s the simple fact, HPE was never going to sell an HPE OpenStack because no one was ever going to buy an HPE OpenStack and HPE has zero f*****g clue how to operate software. But HPE will do a great job at selling servers under a Canonical OpenStack.”
Shuttleworth said that OpenStack will thrive and because of that there will be a need for servers, which HPE, Dell EMC, Cisco and others will provide and build a great business off of it, but some of the big vendors have “taken a big punch to the face,” due to a delusion that they would own the entire stack.
“The delusion that magically they would own the entire stack was what had to fail, so the mixed messages is that they’ve taken a big punch to the face but at the end of the day they will end up selling a bunch of servers into OpenStack and I think that’s great, and they will be successful and we will help them with that,” said Shuttleworth.
“This sounds like tough love but that’s exactly what this is. There is a bunch of delusion behind the vendor strategy.”
Cisco is similar, while the Canonical founder says that his company deploys a lot of its OpenStack on Cisco UCS the vendor along with others have learned that they can’t own the whole stack.
“There’s a difference between a vendor owning the whole thing, which is what they wanted, and participate with things that make sense, which is what will happen.
“In all of that right now is the crisis, but what will emerge is that people will want to buy servers from HPE and they will buy from HPE and they’ll push a button and there Canonical OpenStack will get bigger with those servers and that’s great for HPE.”
In the past, Shuttleworth has, in his own words, “somewhat controversially” called out OpenStack for having lots of things in it that don’t really matter, that can essentially be ignored because they are going to die.
That isn’t necessarily a bad thing for OpenStack because it will help to move it to a place where everyone agrees on what exactly OpenStack is.
The OpenStack Foundation is also seemingly starting to recognise this and has started to move towards wrapping up projects that no one is interested in or using, but it’s been a long time coming and there’s a reason for that – money.
At least that’s according to Shuttleworth who said it is: “because each one of those has a vendor that pays a sponsorship and a subscription.”
But if these projects are closed then will it force companies out and damage the ecosystem? Shuttleworth said that words like ecosystem are associated with winning, hence why they are used so often and that “noise is not associated with winning but they can often be the same thing.”
“So the foundation went through a phase where adding vendors added money and the perception of ecosystem, but it also added noise. So the painful thing that is going to have to happen is that some vendors will have to accept that they fit into a world where the only things that matter are Nova, Cinder, Swift, Neutron, KVM…”
The soon to be CEO said that his point is that it’s none of the vendor specific “stuff.”
“You’re already seeing Mirantis has multiple of those projects they just can’t sustain them, HPE has abandoned multiple of those projects and with that sponsorships will go. That’s ok, it’s the old how many people does it take to change a lightbulb, well if it should only take one then why would you want 20 in a room?”
Shuttleworth doesn’t want to see a single vendor story and urges careful thinking from the Foundation as to how it moves forward without ending up in a single vendor situation, “because they’ve allowed throwing money at the problem to carry a very big stick and there are some people that have money.”
Fundamentally he believes that getting virtual networks, virtual machines, and virtual discs right, doesn’t take 300 vendors.
“We can build ecosystems around OpenStack, very viable ecosystems around OpenStack without making the core contested, that will take real leadership,” said Shuttleworth.
He does feel that the Foundation is getting ready to make the hard decisions but that in the end it will be forced upon them because vendors will leave as they haven’t made money.
“That’s not a bad thing for OpenStack, if you look at what we said three years ago, a lot will go away so don’t fall in love with it. What you should fall in love with is the core that is what will survive and thrive,” he said.
Although it may come across as Shuttleworth being quite doom and gloom about the state of ecosystem, he was actually very positive. His company is doing well, as highlighted by a recent OpenStack user survey that showed Ubuntu as being just behind Red Hat for commercial traction. He’s also seeing increasing traction when it comes to businesses being open to having multiple operating systems – with Ubuntu becoming the new Linux in the building.
While there are positives for Canconical, Shuttleworth still wants to see open source cloud infrastructure’s operations become crisper and more automated in order to get in to line with the public cloud vendors like Amazon Web Services, Microsoft Azure , and Google.
Shuttleworth said: “As long as the operations of OpenStack are crisp and automated the way we do them, that is a beautiful thing. There a lot of people in the OpenStack community who can’t do that, they can’t operate OpenStack, their OpenStack can’t be operated, so operating it is kind of crazy and therefore unpredictable in terms of cost.
“So businesses look at that and see they have a predictable thing from Amazon and this crazy and unpredictable thing from Mirantis, well that’s a shit choice. Whereas if you can get it to the point where Amazon is predictable, Google is predictable and OpenStack is predictable the way it is from Canonical with a managed service like BootStack or with just doing it yourself with MAAS and Juju that is a great thing.”
One thing that could help OpenStack is the upcoming Microsoft Azure Stack which is due to compete with OpenStack when it is fully released.
“That’s an interesting one and I think it’s really good,” said Shuttleworth. “So yes, I think Azure Stack is a serious competitor to OpenStack and I suspect Azure Stack will be good. Historically, version three will be very good and I think that’s good for OpenStack.”
The reason why he thinks it will be good is because it will help to force OpenStack to get rid of the things that don’t matter.
There’s a lot of changes that are going to be on in OpenStack, some necessary, some that will be forced on them by external forces. Increased competition in the market should be a good thing, but it will be up to the Foundation and vendors to make sure they get it right.
Shuttleworth isn’t critical of the Foundation, he said he admires them but that there are things that that they need to learn, not in a way of calling anyone an idiot, but there are a lot of decisions ahead that need to be made.
This article is from the CBROnline archive: some formatting and images may not be present.