Related topics

Hey, biz bods: OpenStack will be worth $3.3bn by 2018

Anything that simple to use has got to be complex to set up

Columns of coins in the cloud

OpenStack looks like it will inject some $3.3bn into the market by 2018, growing from $890m this year according to 451 Research's recent Open Stack Pulse 2014 report.

Much of that (roughly 70 per cent) will be from public cloud providers like Rackspace and HP, but a fair amount will come from private cloud build-out. Unless sentiment changes, companies will be looking to keep cloud safely ensconced on their own enclaves (our surveys*) show companies consistently preferring private and hybrid cloud deployments over purely public cloud. We're seeing strong interest in OpenStack for these private clouds.

The thing is, getting those OpenStack clouds up and running is not as easy as throwing your credit card into AWS, Azure or Google Cloud. To twist the old doctrine: if you build your own cloud, you now have to buy at least a decade's worth of an occupying ground force... or something like that.

Another delightful infrastructure 'adventure'

In my experience talking with companies building their own OpenStack private clouds, there's a growing need for hiring outside help and layering additional products on-top of the private clouds. This is not unlike any piece of core infrastructure in the data centre.

When building their OpenStack program, most companies follow a long cycle of kicking the tires, experimenting with small projects, and then rolling out broader access to the private cloud in stages. The dream is finally creating a self-service pool of resources that developers, big data analysts and others can use without filing ponderous tickets kicking off projects measured in multiple man-months instead of minutes. You know, to get a cloud, except not those pesky public ones.

In this cycle, a company's developers often download an OpenStack distro or source code and stand the clouds up on their own. This usually goes poorly. A common refrain from IT director types at this point is "now my developers are maintaining a cloud instead of writing code".

Finding your 'boots on the ground'

At this stage, services firms come trundling in to help with the systems integration and cloud implementation work. US companies like Mirantis got their start here and still do a little over half of their revenue in services.

eNovace, another OpenStack services firm, was recently purchased by Red Hat for $95m to augment the Linux distro king's consulting abilities in OpenStack. Of course, others like Piston would solve the problem with better software: they say that 95 per cent of their customers didn't need professional services work. But, in the broader OpenStack world professional services work is often needed.

Once the stack is up and running, companies often find the need for additional products with managing and monitoring the new private cloud as an obvious add-on. Forcing companies to tweak their service delivery processes to move towards a more DevOps-oriented approach is another soon-to-be-dire need: if you spend all that time and money building your own cloud to achieve faster time to market and then follow the same old slow-to-market processes, you're doing it wrong.

Swapping out that core engine of infrastructure to a faster, cloud-driven core elicits all sorts of utopic promises. One of them isn't the complete removal of the need for outside help to figure out how to customise all that generic, cloudy awesomeness into the specific day-to-day infrastructure drudgery that drives a business's profits. ®

* Michael Cote heads up the Infrastructure Software practice at 451 Research

Sponsored: 5 critical considerations for enterprise cloud backup