OpenStack at 10 years old: A failure on its own terms, a success in its own niche
And with China’s web giants showing it is capable of enormous scale, there’s still plenty of potential
Analysis When OpenStack announced itself on July 19, 2010, it gave itself a mission “to produce the ubiquitous OpenSource Cloud Computing platform that will meet the needs of public and private clouds regardless of size, by being simple to implement and massively scalable.”
Measured against that mission statement, the project has mostly failed. OpenStack has proven massively scalable, but hyperscale public clouds don't use it. It is also far from ubiquitous – analyst firm Gartner can find around 2,000 serious production implementations. Meanwhile, tens of thousands of private clouds run hyperconverged infrastructure stacks from either VMware or Nutanix, strongly suggesting they have cornered the market among buyers who value simplicity. VMware may also rule the public cloud, as over 4,000 clouds run by service providers employ Virtzilla’s software.
OpenStack’s aspirations didn’t seem unreasonable in 2010, when NASA and Rackspace formed the project and quickly built a community that aimed to create an open-source cloud platform. That community delivered its first release – named Austin – on October 21, 2010.
“It had a massive agenda which was to counter Amazon,” said Michael Warrilow, a Gartner research vice president.
Tristan Goode, the CEO of OpenStack consultancy Aptira and a former member of the OpenStack board, recalls an early OpenStack event in Beijing during which a Chinese government speaker said: “OpenStack will smash the monopoly of the Western cloud providers”.
OpenStack's 10th birthday is next week, but you get the present of a new release todayREAD MORE
And for a while that agenda advanced nicely. Rackspace led the charge with a spiel that proprietary clouds create the risk of lock-in while open clouds represent less risk and faster innovation. HP and Cisco created their own OpenStack-powered clouds and used similar arguments.
Charlie Dai, principal analyst at Forrester, thinks things went well in the project’s early years, at least when measured against other open source projects. “In the first five years OpenStack won the war of resource-oriented infrastructure management against CloudStack and Eucalyptus and became one of the most appealing opensource cloud platforms for both on-premises and vendor-hosted services,” he said.
While adoption outside of telcos and China was not enormous, the project continued to attract developers and achieved a steady two-releases-a-year cadence. However, as new modules were added to the project, simplicity proved elusive and it gained a reputation for complexity.
OpenStack private clouds are SCIENCE PROJECTS says GartnerREAD MORE
Aptira's Goode says the increased interest in the project also created some conflict. “I think the hype kind of killed the mission because it became a tug of war between vendors who were rapidly seeing their businesses disappearing to the major clouds,” he told The Register. Vendors’ ambition to protect their own business “pulled it away from being a cloud platform into being an infrastructure platform,” he said.
That shift did not go down well.
“I remember the horror of running things other than cloud workloads was real,” Goode recalled. Battles were fought over features such as live migration of virtual machines because they weren’t considered part of the original OpenStack vision.
Gartner's Warrilow also observed a loss of focus. “OpenStack splintered because there were different commercial incentives,” he said. While OpenStack fought its battles, buyers just kept sending cash and workloads to AWS, Azure and other hyperscale clouds.
Warrilow thinks that happened because they offered the simplicity OpenStack hadn’t been able to deliver. “Everyone says they want open, then they say they want easy,” he said. “They end up in the middle with an easy closed solution.”
Big backers of OpenStack started to realise that customers weren’t interested. HP killed its Helion cloud in 2015. Cisco’s Intercloud was shut down in 2017, having lasted just three years. Rackspace “pivoted” from OpenStack hosting and services to become an organisation that offered support for any cloud.
But while clouds based on OpenStack foundered, China’s big users showed it could power services at colossal scale. Tencent and Baidu led the way, each scaling to over a billion users on OpenStack infrastructure.
Yet those demonstrations of OpenStack’s scalability generated little interest among other potential users. Warrilow said Gartner now receives hardly any inquiries from clients about OpenStack. And you don’t need to be an analyst to observe the massive growth of AWS, Azure, Google Cloud, and other hyperscale cloud players.
Goode said OpenStack did find an important niche as an open and customisable cloud, which matters to telcos and government users.
OIF! Forget OpenStack Foundation, it's becoming Open Infrastructure Foundation - with new members, an expanded roleREAD MORE
“I am a little bit biased as my business depends on OpenStack, but there is a valid uses case in terms of its customisability,” he said. “Users with particular security concerns have no other choice” when shopping for a private or hybrid cloud, he said. “There is a need for air-gapped networks.” That the project remains open, and therefore extensible, also means some users will continue to engage and adopt the stack.
Forrester’s Charlie Dai also thinks the project has a future.
“Looking into the future of cloud, decision-makers and practitioners are strategically shifting their focus from resource management on infrastructure layer to application orchestration on full-stack for technology adaptiveness,” he told The Register by email.
OpenStack is meeting those needs with both its code and its structure.
“The latest announcement of OpenStack Foundation into Open Infrastructure Foundation is a smart move, helping OpenStack stay relevant in the cloud space by focusing on the infrastructure support in the hybrid environments across cloud and edge with improved support for K8s, service mesh, CI/CD and other cloud-native technologies.”
Today, OpenStack has a new mission.
The project now aspires to “produce a ubiquitous Open Source Cloud Computing platform that is easy to use, simple to implement, interoperable between deployments, works well at all scales, and meets the needs of users and operators of both public and private clouds.”
And that mission appears to have found a new set of admirers to turn it into reality: Gartner’s Warrilow said Chinese hyperconverged infrastructure vendors are mostly building on OpenStack. Even if their efforts aren't big hits outside China, the prospect of Chinese SMEs adopting on-prem private clouds running OpenStack and extending them into hybrid clouds that also use the open source project would deliver on much of the current and initial mission statements. ®