If the open source model is broken, as Apcera founder Derek Collison believes, then container orchestration wunderkind Kubernetes may be its first major casualty. Yes, that Kubernetes, the Google-spawned container king that 71 percent of enterprises surveyed by 451 Research say they’re using for container management.
It seem far-fetched that Kubernetes could be heading for a fall, even as it continues to rise. But the problem, Collison argues, is one of investment: The old open source model was all about commoditizing a richly funded market filled with proprietary software. Open source came along, democratized the market, and shifted investment dollars elsewhere.
Kubernetes (and a host of other new-school open source projects) turns this model on its head, however. It started life as open source and thereby rendered the initial price as $0.00. This is OK for now while Google funds Kubernetes to provide an API-laden path to its cloud, but what happens if Google finds more efficient ways to drive paying customers to the Google Cloud Platform and other services?
Collison believes Kubernetes is setting itself up to be the OpenStack of container orchestration (that’s not a good thing). Others from Red Hat and CoreOS, however, tell a different story.
The open source of yore needed proprietary software to commoditize
It would be easier to dismiss Collison’s view if he hadn’t been around the industry for so long, and in such influential positions. Collison cofounded the AJAX APIs group at Google, where his experience with Google’s internal Borg led him to architect CloudFoundry while at VMware. Now at Apcera, an enterprise container management platform, it’s fair to suggest Collison may have an axe to grind against open source alternatives like Apache Mesos and Kubernetes.
It’s also true that he has the requisite experience to grind that axe with some accuracy.
Collison isn’t an anti-open-source crusader. Far from it. As he tells me in an interview, “Open source is a natural progression for ecosystems where there’s a lot of innovation and breakthroughs. As markets mature, they eventually become democratized and open source alternatives emerge.” Individual vendors may choose to ignore or fight this natural evolution (as Microsoft once did with Linux), but the market tends to get things right, he opines.
But even this natural evolution sometimes goes awry. Pointing to OpenStack, Collison argues it failed despite trying to follow this path of democratizing a market (mostly VMware), perhaps because OpenStack tried too hard to be all things to all people. As Collison argues, OpenStack had a governance model but couldn’t govern the warring fiefdoms that tried to take it in multiple directions, with a crazyquilt of tech options to accomplish the same thing. Although some companies, most notably Red Hat, have managed to build viable businesses from OpenStack, the startups around it were wiped out, earning professional services revenue but little product revenue.
In short, Collison tells me, the perceived value of open source OpenStack is “zero,” making it impossible for an ecosystem to flourish around it.
Maybe open source is also about value creation
Joe Fernandes, senior director of OpenShift product management at Red Hat, rebuts the claim that open source is best served as a commoditization tool: “Open source is a critical source of innovation, not just commoditization, and it is at the core of many businesses, including software vendors, cloud service providers, and telcos.”
Even so, Fernandes’s colleague Brian Gracely, director of OpenShift product strategy, piles on, arguing, “Kubernetes is actually trying to commoditize extremely valuable (and proprietary) technology: Google Borg.” Although not strictly true (Collison is deeply familiar with Borg and told me “Google isn’t open-sourcing the Borg tech” with Kubernetes), it is true that Kubernetes represents the ideas that shaped Borg, and are being made generally available through open source.
Indeed, Fernandes continues, Kubernetes is a fantastic example of open source’s renewed vitality, one that is helping different businesses to thrive:
The Kubernetes project is thriving and is one of the fastest-growing open source projects in history. Although Google initiated the project and Red Hat is a top-two contributor, Kubernetes has a large and growing ecosystem of contributors that includes companies both large and small, including CoreOS, Heptio, Rancher, IBM, and Microsoft. These companies are investing in Kubernetes precisely because they see that it is a huge opportunity to advance their business interests.
Collison, however, isn’t so sure that these disparate interests can be made to build a coherent project. Remember OpenStack?
Cash for Kubernetes? You’re kidding, right?
Kubernetes differs from OpenStack in its provenance, Collison tells me, but “it’s repeating some of the mistakes that OpenStack made, while also not acknowledging that customers’ perceived value of it is zero.” Of course, vendors will make money around Kubernetes, he acknowledges, but “not product revenue—only professional services revenue.” Red Hat and others may beg to differ (at time of publication I had requested but not received comment from others), because Red Hat’s OpenShift is based on Kubernetes and is the company’s emerging cash cow.
But Collison offers a more pointed critique: What if Google, the top contributor to Kubernetes, bails on it?
The only reason Kubernetes exists, Collison declares, is “so Google can open-source the API ecosystem to drive workloads to its cloud.” If that strategy proves impotent, or if Google finds a superior way to drive Google Cloud Platform adoption, “would Google still care about Kubernetes at that point? Likely no.”
Red Hat has a solid financial incentive to continue to invest in Kubernetes, and new entrants (like Oracle) could further foster that investment. But without a Linus Torvalds-type keeping the tech honest, it’s plausible to see Kubernetes repeating OpenStack’s fracturing.
That concern, Fernandes tells me, is misplaced in light of the actual governance of the project. “The Kubernetes project has an open governance model that is based on meritocracy,” he says. “Leaders from Red Hat, like Clayton Coleman, join leaders from Google, like Brian Grant and Tim Thockin, as well as leaders from other contributing companies, to shape the direction of Kubernetes and associated Special Interest Group projects.” Far from being beholden to Google, he stresses, “Kubernetes is not dependent on one vendor and is not slowing down. Rather, it has a sizable and thriving community and it is accelerating the pace of container innovation.”
CoreOS founder Alex Polvi agrees, telling me that “Kubernetes has won in part because of community—no one company can kill it. It’s a true vendor-neutral ecosystem.”
Ironically, one of the biggest areas of vendor contribution may end up coming from the public cloud vendors, notes Red Hat’s Gracely, “because hybrid cloud will be the dominant buying model for the next decade.” That calls out the need to ensure that Kubernetes serves not only as an API gateway to Google Cloud Platform, but also an on-ramp/off-ramp for Microsoft Azure and Amazon Web Services. In fact, as CoreOS’s Polvi points out, “Look closely and you’ll see AWS in the Kubernetes community already.” This is particularly significant for AWS because, he stresses, “Amazon doesn’t like to waste time on open source development.”
So far, the rosier, inclusive view of Fernandes, Gracely, and Polvi seems to be playing out. But Collison’s concerns shouldn’t go unheeded. Google didn’t open-source Kubernetes out of charity, and its presence lends engineering muscle to a project that increasingly can stand on its own but may not yet return enough financial return to thrive in its absence.
Fortunately for the Kubernetes faithful, Red Hat, CoreOS, and others are increasingly picking up the slack, if any, that Google may be leaving. They have to. After all, there were container-orchestration solutions before Kubernetes, but none with the depth and elegance it provides. As enterprise vendors push innovation rather than merely IT cost-cutting, Kubernetes is their Google gift that promises to keep on giving.