As I winged my way back home after attending OpenStack’s Summit in Barcelona last week, I was struck by a growing suspicion that the event marked a turning point for the open source cloud initiative, and that its future is going to start looking different to its past.
I’ve been following OpenStack since its inception six years ago. Back then the project, initially conceived by Rackspace and NASA, saw interest from a couple of different areas.
Firstly there were a high number of startups which were founded with an aim of commercializing OpenStack. These startups, largely founded by people had been involved with the project either at Rackspace or NASA, secured copious amount of early-stage venture capital. Most of these companies have folded or been acquired (and generally subsequently shuttered) -- Nimbula, Piston Cloud and CloudScaling are now consigned, in part or in whole, to history.
The other big class or organizations jumping on OpenStack were the legacy enterprise vendors. These vendors had spent the previous five years disputing cloud as a way of delivering infrastructure and saw the writing on the wall -- the world was inexorably moving towards cloud and they needed to find themselves a cloud story.
These vendors jumped on board OpenStack, both with their own resourced development teams, and through acquisition -- IBM, HP, EMC and Cisco all had huge presence at previous OpenStack Summits.
Walking around the show floor last week, however, things felt significantly different. HPE, the company formerly known as HP, is a shadow of its former self and has had a seemingly never-ending serious of missteps and pivots. HPE was almost invisible on the expo floor and, other than a private briefing I had with them, their presence at the Summit was invisible. Scuttlebutt around the summit was that the company have fired almost the entire team working on both its OpenStack distribution and Stackato, its Cloud Foundry-based platform-as-a-service offering.
Ditto for Cisco, the company that in previous years had a strong OpenStack advocate and figurehead in executive Lew Tucker. Cisco’s presence on the expo floor was minimal and there was no indication that OpenStack is a focus for them anymore. As for Tucker, in years gone by he would have been front and center of the summit, but I saw neither hide nor hair of him. EMC were in attendance, but their obvious focus on the recent merger with Dell has taken their eye off the ball -- or at least that’s the way it looked in Barcelona.
Of the four legacy horseman of the OpenStack ecosystem, only IBM had a significant presence and even theirs felt, to this commentator anyway, a little more muted than in previous years. And Mirantis, the pure play OpenStack vendor that has always punched above its weight in terms of impact at the show, was also a little quieter than in previous years.
It would be easy to suggest that this rapid change in the ecosystem is an indication that the world has moved on from caring about virtualization platforms and is instead focusing on newer, more exciting stuff -- things like Docker, Kubernetes and the like.
But that ignores the traction that we saw at the show. The reality isn’t like the cool cloud-native and containerized kids would have you believe -- the general consensus among the analysts chatting at the show was that, while all things containers are top of mind for vendors and the Silicon Valley cool kids, real enterprises are more conservative in their adoption of these newer technologies.
Indeed the day one keynote saw an absolute barrage of user case studies that came from commercial and non-commercial organizations and from a variety of industries. There was perennial show favorite, Tim Bell from CERN talking about the sheer scale of their OpenStack infrastructure (close to 200,000 cores in production) and a representative from the Square Kilometer Array, an ambitious project building the largest space telescope in the world. The SKA has almost imaginable scale, it will produce around 5,000 PB of data every two days, all of which is going to run on OpenStack clouds.
And beyond the science community, there were some impressive commercial users. In an NDA analyst-only session we heard from one massive retail brand who has 60 percent of its total IT footprint and close to 100 percent of its externally facing web assets on OpenStack. And telco’s, for a few years now seen as one of the bigger commercial opportunities for OpenStack, were there en masse.
I spent time talking with Zhang Zhihong, deputy general manager of cloud computing products at China Mobile, China’s largest telco with 835M subscribers. The company, a newly minted OpenStack corporate sponsor, has its own OpenStack cloud with around 11,000 servers across public and private cloud offerings. China Mobile also has plans to expand that by another 6,000 servers imminently. OpenStack Foundation executive director, Jonathan Bryce, put it into human (albeit Texas) terms when he told the audience that China Telecom plans to deploy the equivalent of 500 acres of OpenStack-powered data centers. Scale almost unimaginable.
But here’s the thing -- CERN and SKA, while amazing examples of the success and production-readiness of OpenStack, are scientific projects built in-house. None of those commercial vendors on the show floor make much money from them. The retail case study we heard about is built largely by an inside team, and China Mobile has its own OpenStack team that contributes back to the project.
Four very different end users of OpenStack, and all impressive in their own right, but none of them making money for commercial vendors.
In a report commissioned by the OpenStack Foundation, and written by 451 researchers, much mention is made of the growing scale and production focus of OpenStack deployments -- about 72 percent of OpenStack-based clouds are between 1,000 and 10,000 cores. The findings were interesting -- production readiness and a move beyond simply test and development, a range of different industries and at last the long-heralded maturity. All of which is a fantastic vote of confidence for the project, but not for the product and service companies built on top of it. Perhaps partly the reason that big vendors, many of whom have invested millions of dollars in their OpenStack initiatives, seem to be retrenching somewhat.
In an interview at the show, Canonical founder, creator of Ubuntu and spaceman Mark Shuttleworth struck an emotive note when he suggested that the OpenStack project should dispense with producing, in his words, “bull---- as a service.” Shuttleworth was referring to the very real issues of multiple parties developing different add-ons to OpenStack, rather than focusing on the all important fundamentals of compute, storage and networking. But behind Shuttleworth’s stylized BS as a service lies vendors who have not a clue about how to actually make money from OpenStack -- that is the driver behind the perennial feature sprawl that has plagued OpenStack.
In one way it is a relief and a real source of satisfaction for everyone concerned that we no longer need to write the obligatory “is OpenStack a thing” post after a summit -- OpenStack is here, it is stable and it is delivering real world value to end users. Whether OpenStack will generate significant value for the commercial vendors, however, is another thing entirely.
And for the answer to that question, we will have to wait.
This article is published as part of the IDG Contributor Network. Want to Join?