The cloud is expanding, with more services coming online from established providers and new entrants. But each time a major player such as Amazon Web Services or Rackspace is marked out for its potential to expand its cloud portfolio, it raises the question of how open the marketplace will be for customers if a handful of big providers end up dominating the industry.
Neil Levine has headed up corporate services at open source Ubuntu-backer Canonical since August 2009. Prior to that, he was the chief technology officer at Claranet, which provides managed hosting from the datacentre up to the application layer. Enthusiastic about open-source schemes within the cloud, he recently attended the launch of the OpenStack scheme, an attempt to make it easier for companies to get off the ground in cloud computing.
Levine spoke to ZDNet UK to discuss the role open source has to play in keeping the barriers down for companies providing services within the cloud, and how the move to enterprise cloud computing is developer-led.
Q: Mimecast's Nathaniel Borenstein told us he thought the cloud in 2020 could be dominated by a handful of large players. What do you envisage: a cloud dominated by major people with platform lock-in or something based more around a variety of smaller companies with more distinct purposes?
A: I'd be surprised if it was just three majors — I think companies won't like that. But then there are economies of scale — tremendous ones with someone like Amazon — so it's going to be somewhere in-between. I'm sure you will have some islands of people and there will be lock-in, but it will be because there will be great technology involved.
What we want to see is open standards and open source — no company controls the technology. This is interesting about OpenStack: it's not just the software development, it's that the design around the cloud stack is open as well. Some big people may have lock-in, but we would hope to see some real perfect competition here. Being able to move between the companies where the technology, implementation and application programming interface [API] are the same and software may or may not be the same and the standards are the same — it's a service definition.
Why is it important for OpenStack to be compatible with Amazon's Elastic Compute Cloud (EC2)? Is it because of the ability to migrate data or the ability to interface with it?
Migration, because if you're a company and [Amazon Web Services] suddenly jack up its prices, then you'll be able to move off. Banks are a good analogy here — I can move my money from one bank to the other and I don't need to worry about what accounts I am moving into; they may have different saving products, but I can still move my money around. So, the cloud in 2020 hopefully will look more like a mature utility market where there is service differentiation and there is commonality to allow portability. But yes, there will still be some traditional big, big vendors who through some particular cost-advantage or just size, form a big gorilla in the space.
Fabio Torlini, Rackspace's head of cloud for EMEA, said that the company could gain competitive advantage by being early members of the OpenStack scheme. Is that the incentive for companies being part of open source and open standards in the cloud?
Definitely. I think Rackspace's point of view is that cool kids are receptive to cool messages. It also doesn't want Amazon to be driving this industry and be out in the front. If it can be associated...
...with a source of innovation that it has involvement in and is associated with the project before Amazon, then it'll get some of that mindshare around it. So I think all these reasons bolster each other.
I don't think OpenStack will be the only game in town. Distributed Management Task Force [DMTF] has something like six or seven standards being submitted to it currently so we'll see how Amazon responds. There's a little bit of an echo chamber going on at the moment: you have to see what the software is like, who adopts it, who uses it. There's a lot still to happen, but [Rackspace] has done it for the right reasons from its point of view and I think it has set up the project in the right way.
Is your contention that, because of how innovative the cloud is becoming and how fast the field is moving, it's becoming less about the platform and more about the software that is mounted on it?
Exactly, absolutely. Because everyone is trying to work out what the coolest piece of infrastructure they can be running is — whether it's Eucalyptus or OpenStack — and then what kind of workloads they want to run. You're seeing this big data meme where people want to start using Hadoop or Cassandra or these other NoSQL technologies and we get them [Canonical's Ubuntu releases] out there every six months — this is a developer-led shift here. The operations people are billing their needs, but ultimately it's the application people who are saying: "This makes my job easier; I can write better applications that scale more easily if you give me the infrastructure".
So you would say it's a developer-led process?
In the enterprise, yes. We've had a product on the market for over 1.5 years, and we were the first open source cloud option. We took the product to a lot of businesses and they talked to their operations people and infrastructure people who said: "yeah, we love this, this is great, but I can't just deploy it with nothing to run on top. I have to have an application". Then they talked to their application people who said: "I want to do cloud, but this is the problem I've got at the moment" — whether it's analytics or big data issues.
That's why a lot of people are using Amazon — they've busted out of the firewall because they couldn't get enough of the stuff from their operations people, so it's the application people who are catalysing these decisions. By getting [Ubuntu] out every six months with the latest cloud software in it, we're giving the developers what they need to say to the operations people: "Give me Ubuntu so I can roll this thing I was playing with out".