Unpacking the OpenStack Foundation’s open infrastructure push

Embracing open infrastructure

“New demands are being placed on IT infrastructure with emerging use cases that are growing beyond what was initially imagined and workloads that are no longer contained to central datacentres,” said the OSF in an open infrastructure strategy document circulated at the summit.

“The OSF recognises that no single technology solution is going to support this transition and the integration and knowledge-sharing around these open technologies is key to successful implementation.”

The latter part of this statement concerns the interoperability issues the OSF has previously flagged as impediments to getting enterprises to wholeheartedly embrace open source, as the offerings produced by one supporting community may not necessarily work well together natively.

Overcoming these integration challenges has seen OSF advocate for adjacent open source communities to work more closely and collaboratively, as building out an infrastructure stack with open source technologies should inherently be easier than trying to achieve the same result by adding proprietary technologies into the mix, too.

“When you pull OpenStack, there are a lot of other pieces you need [to make it work] from networking drivers, storage drivers, compute drivers,” Alan Clark, OSF board chairman and director of industry initiatives at OpenStack distribution provider SUSE, told Computer Weekly. “And that is why, to me, it makes sense to talk about open infrastructure.

“If you try to mix [those technologies’] proprietary technology with open source, you start to find impedance barriers, whereas if you push for an open source stack, it’s a lot easier to pull those pieces together.”

There are also productivity benefits to be had from running entirely on open technologies, said OSF executive director Jonathan Bryce during the opening keynote at the OpenStack Summit in Berlin.

“When an organisation is able to take those [open source] tools, integrate them successfully and build them, it creates big opportunities for them,” he said.   

“One of the biggest is the ability to innovate faster – in some cases, faster than the market – and that can give them a competitive advantage.”

Clark expanded on this statement during a follow-up interview at the summit by pointing out the considerably larger pool of engineering talent that open source projects have got to draw on during their development, compared with proprietary technologies.  

“The OpenStack Foundation is operating in 187 countries, and it would be very tough for a proprietary vendor to reach that kind of scale and to dominate,” he said.

“At SUSE, we have engineers across the world, and because it is open source, it is easy for us to bring people on, integrate them and make than an integral part of the company, so we can pick up employee talent working anywhere in the world.”

From an OpenStack perspective, access to such a skills base has culminated in 70,000 updates being made to its core cloud platform in the past 12 months alone, said Clark, making it one of the world’s top three most active open source communities.

“If you talk about from the business side down, they may not care that it is open source [that their infrastructure is running on], because what they’re really striving for is agility, cost containment, rapid ease of deployment and a faster route to market,” he said. “And those business reasons are all strengths of open source.

“Leveraging all that innovative effort in open source means enterprises can get to market quicker and beat the competition. Plus I get to take advantage of all the talent – not just regional or local, or where I happen to have an office, but globally.”

Clearing up OpenStack cloud confusion

As previously documented by Computer Weekly, there remains a degree of confusion among enterprises about what OpenStack is and what technologies make it up, and this is something the OSF has publicly declared its commitment to address in recent years.

According to Nick Barcet, senior director of OpenStack product management at Red Hat, open infrastructure messaging also plays a role here.

“The shift towards open infrastructure is, I think, associated with those misconceptions, where people have thought of OpenStack as a virtualisation management platform, and it was never just a virtualisation management platform,” Barcet told Computer Weekly.

“At the beginning, it was what people knew it for, so they made that association, but now we need to disassociate from that and make it easier.”

The OSF’s efforts to clarify what OpenStack is has also involved winding down the Big Tent governance model, which was ushered in during May 2015 in an effort to redefine how OpenStack projects should be categorised.

Whereas previously, contributors had to petition to have their projects included in a future OpenStack release before they could start work on them, Big Tent effectively saw this administration barrier lifted. Instead, as long as the projects they wanted to work on stuck to the OSF’s guidelines, contributors were good to go.

An unintended consequence of all this is that lots of new projects and add-ons emerged for the core OpenStack platform that were of negligible use or interest to users, or lacked the support of the wider contributor community. This, in turn, left enterprises confused as to what OpenStack is really about.  

Canonical founder Mark Shuttleworth, whose company makes the Ubuntu distribution of OpenStack, told Computer Weekly there is a risk that, having abolished Big Tent, the open infrastructure message could lead to further confusion.

Particularly as the broadening of the OSF’s focus means the OpenStack platform is now far from the only offering its 82,000-strong community of supporters can contribute to, as the past 12 months have seen the organisation debut four additional pilot projects it is supporting.

These include Airship, a cloud provisioning and lifecycle management tool designed to help enterprises deploy and manage containers, virtual machines and bare metal infrastructure environments across multiple sites, including edge computing environments.

The project, and its resulting technology, is the result of a collaborative effort between South Korea Telecom, Intel and telecoms giant AT&T, with the latter revealing details during the summit of how Airship and OpenStack platform is underpinning its 5G network.

The StarlingX platform is another of these pilot projects, which the OSF bills as a cloud infrastructure software stack that can enable organisations to stand up applications in distributed edge environments, and whose version 1.0 release made its debut in late October 2018.

There is also Zuul, a continuous integration and delivery platform that is in use at BMW and French e-commerce company Leboncoin, and is already on its version 3.0 release.

Perhaps the best-known of these four pilots is Kata Containers, whose version 1.0 release coincided with the OpenStack Vancouver summit. Its aim is to create standardised, lightweight virtual machines that can be used by enterprises concerned about the workload isolation and security risks associated with using containers.

“The end of the Big Tent was a big improvement and it mainly came about because most of the vendors [supporting those projects] failed to get any big traction with OpenStack, so they went away and so did their noise,” said Shuttleworth.

“The question really is whether the foundation is driving these four projects based on what users want or what vendors are paying the foundation to publicise.”

Barcet said some of the projects are more interesting than others to Red Hat. Chief among them is Zuul, given that the company is already a keen user of the technology and a number of its employees are major contributors to the project.

As for how involved it plans to get in the gestation of the other three projects will be depend on a number of factors, including whether or not there are technologies out there doing the same thing, how much demand there is, and the quality of work going into them.

“We have to be very picky because projects with a single contributor or that are stagnating would become a big liability for us, as there is a risk they will not be sustainable over time,” said Barcet.

“We are not always right, but at least we are less often wrong by doing this analysis before we support something.

“Overall, we try to be pragmatic in our approach to open source projects and not jump to the hype du jour.”

Keeping focus

The emergence of these pilot projects coincides with the creation of the OSF’s Strategic Focus Areas initiative in December 2017, the intention of which is to focus the minds of the foundation’s supporters and contributors on addressing common technology problems in edge computing environments, containers and datacentres.

From 2019, the foundation will add artificial intelligence and machine learning to its list of Strategic Focus Areas, which, in turn, could give rise to one or more additional pilot projects.

But just because the technology areas the foundation supports are expanding, that does not mean its commitment to the OpenStack platform is being wound down at all.

Quite the contrary, in fact. During the summit, the OSF repeatedly referred to the platform as being the cornerstone of its open infrastructure strategy.

Even so, OSF chief operating officer Mark Collier said the pace of innovation each of these four initiatives is already bearing witness to suggests the foundation is on to a winner.

“As the governance structure of the OSF is evolving to meet the needs of our open infrastructure community, these four pilot projects are already demonstrating impressive progress,” he said.

“What they’ve accomplished in a short period of time is a proof point that focusing on open infrastructure is the right model, putting users at the centre of everything we do.”