Can the platform shrug off the problems of the past and find a place inside more enterprises?
OpenStack has been heralded as being on a trajectory to "win" the race to become the platform of choice for businesses building private clouds.
But the rate at which enterprise has adopted the cloud controller platform has not met the expectations of some. When OpenStack start-up Nebula shut its doors this month, it indicated the slow pace of adoption of OpenStack had played a role in its demise, saying "we are deeply disappointed that the market will likely take another several years to mature".
The problems that make businesses reluctant to deploy OpenStack are nothing new. For several years, leading voices in the OpenStack community have been warning the project has been too focused on adding new features instead of stabilising its core.
The unreliability of certain core projects, specifically the Neutron networking project, has reportedly led to situations where firms have spent tens of millions trying to build OpenStack private clouds "that will return little value".
"It's not in any sense, especially on the networking side, a completed, battle-tested product ready to ship. It is a work very much in progress. The networking component of it, Neutron, is a good example of that," said Carl Brooks, service providers analyst with 451 Research.
"Most implementations of OpenStack that I'm familiar with still use commercial networking product in place of OpenStack's network capabilities."
Given the complex mesh of legacy systems that exist within most long-established businesses, OpenStack's unpredictability makes it too much of a gamble for many to try to integrate, said Brooks.
"It's a barrier to integration into the enterprise or the IT organisation. It doesn't permeate enterprise and become the broader technological foundation because it doesn't always work as advertised and it takes technical acumen to make it sit up and roll over."
OpenStack was an open-source project started in 2010 by Rackspace and NASA and today is backed by more than 200 vendors, including Cisco, Dell, HP, IBM, Intel, Oracle, Red Hat, and Vmware. Members contribute funding and code to the project, with platinum sponsors each donating $500,000 a year. Altogether the project has an annual budget of around $5m.
"Given the size of the investment that has been poured into it over the past five years, it's certainly not where people expected it to be," according to Gartner research VP and distinguished analyst Lydia Leong.
"There's a lot of experimentation at an early stage, some things for lightweight development environments, but there's not a whole lot of OpenStack adoption at this stage."
OpenStack has a large developer community working on a range of loosely-coupled projects, with components orchestrating virtualised compute, storage and networking infrastructure at its core.
Leong attributes the technical difficulties in part to the way the project is managed. Those contributing code to the open-source platform hold an array of differing opinions as to what OpenStack should be, she said.
"I think that doing open source work in a full committee style is often like pouring 1,000 engineers into a barrel and hoping they'll produce the works of Shakespeare. The monkeys in the barrel just don't manage to get it together, everybody wants to be the king and the directions and the priorities change.
"It's a very different situation to something like Linux, where you have a benevolent dictator Linus Torvalds controlling everything, or like Docker, where there is a corporate entity ultimately controlling the road map."
The controlling OpenStack Foundation encompasses both large service providers like Rackspace, which builds OpenStack private clouds for itself and customers, and major technology vendors like Intel. Within the foundation, there has always been tension between the wants of different OpenStack users, according to Brooks.
"As OpenStack continues to develop it's going to have that many-headed hydra future for a while because there are so many people trying to stick what they like into it."
To increase adoption, the OpenStack Foundation has also been urged to standardise the interfaces that developers use to automate deployment of OpenStack cloud infrastructure around those used on Amazon Web Services, the world's largest public cloud platform. In 2013, OpenStack Foundation founding board member Randy Bias said "embracing of Amazon will put OpenStack in the pole position to dominate hybrid cloud".
Some progress has been made on AWS compatibility since Bias spoke out, with the recent release of a set of APIs for OpenStack's Nova compute fabric controller that are compatible with those used with AWS Elastic Compute Cloud (EC2).
Even though reliability of the core OpenStack projects has improved in the past year, with the formation of a stability team to tackle these issues, Gartner's Leong believes adoption will be hampered by business apathy towards private clouds.
"They've [OpenStack members] been striving to work towards a product that can be consumed and bought. But it hasn't taken off. It's coinciding with the general struggles private cloud has within the enterprise."
In coming years, in place of building private clouds, Gartner believes businesses may instead choose virtual automation. This practice sees firms apply limited automation to the management and deployment of their virtualised infrastructure, rather than run a full cloud management platform.
"Most organisations are backing off private cloud in favour of what we consider to be more virtual automation, taking their existing virtualisation and adding more automation and a little bit of self service without going to full private clouds."
Brooks is slightly more optimistic about OpenStack's future, expressing a belief that the core parts of OpenStack - Nova handling virtualised compute, Swift object storage, Cinder block storage and Neutron networking will become easier to work with and take-up will increase.
"Those things will continue to improve at a metronomic pace," he said.
Even though enterprise use of OpenStack isn't where some had hoped it would be, there have been notable successes, such as its use by payment provider PayPal, and Brooks says deployment is picking up.
"We do see that it's accelerating in terms of adoption. From 2013 until the end of 2014 we saw the number of OpenStack deployments at enterprises essentially double."
In the near future, however, Brooks said spending on OpenStack-related products and services will remain a fraction of the wider $100bn cloud market - with 451 Research predicting they will generate more than $1.7bn revenue by the end of 2016.
"In terms of the overall cloud market it's a fairly small portion of it."
The OpenStack Foundation had not responded to a request for comment at the time of publication.