First things first
I'll get to Kubernetes, but first: Have you seen this meeting taking place today at the White House, featuring a bunch AI researchers, industry execs and government officials? We have to start somewhere, but something about this seems off.
Perhaps it's the focus on "removing barriers to its application" rather than on figuring out ways to strategically apply AI to solve specific problems. As I've been railing about for the past couple weeks, there are real problems in the United States and elsewhere that won't be magically solved by waving an AI wand at them. In fact, there are a lot of problems -- cybersecurity being perhaps chief among them -- that might actually be exacerbated by AI.
My hope would be for real, frank discussion about problems and opportunities, and concrete cross-institutional efforts for addressing them. Facebook's VP of AI will be there: Remember a few weeks ago, when Mark Zuckerberg was getting "grilled" before Congress? A meeting of execs acting as lobbyists does not instill confidence. But independent voices don't make headlines.
The myth of the Kubernetes ecosystem
On the heels of last week's KubeCon event, TechCrunch and VentureBeat each published a profile of the Kubernetes movement:
- As Kubernetes grows, a startup ecosystem develops in its wake (TechCrunch)
- Kubernetes and microservices: A developers’ movement to make the web faster, stable, and more open (VentureBeat)
They're good articles addressing what it undeniably one of the biggest movements in enterprise IT since the advent of the cloud, but I think it's fair to ask at this point what the actual opportunity is for startups in the Kubernetes space. The microservices space, as well as the general developer-tool space, is filled with startups that definitely touch or support Kubernetes, but Kubernetes itself seems mostly like a business opportunity for the big boys.
With few exceptions (Heptio springs to mind), direct Kubernetes development and definitely revenue appear driven my a handful of companies: Google, Microsoft, Red Hat and even IBM. Amazon Web Services hasn't been particularly active in the community, from what I can tell, but it also will make a nice chunk of change selling Kubernetes flavors of its various container-orchestration services. The newly IPOed Pivotal is also doing work to integrate Kubernetes with Cloud Foundry, which has an enterprise footprint that would make many projects envious.
Even large software vendors like Oracle and Cisco are building out their Kubernetes product lines and businesses, because there's money to be made even if you're not a cloud-native thought leader.
If there are "startups" poised to do a lot of business here, they might be companies like Docker and Mesosphere. Neither of them began life Kubernetes companies, but they've both raised hundreds of millions of dollars (see below re: Mesosphere) and have been able to surf the wave and end with Kubernetes distributions.
This isn't an indictment against the Kubernetes project, any of the companies involved or even the CNCF, which houses the project. If anything, it's a testament to the status of open source software in 2018 and magnetic pull of Kubernetes, in particular. Just 4 years since it was announced, big end-users already want to make Kubernetes a focal point of their infrastructure and development practices going forward, and they want to work with large vendors to help them pull it off.
There's definitely a place for startups in this world, but they need to be clever in finding ways to make Kubernetes work for them. The open source project is moving so fast that what seems like a hole-in-need-of-filling today might very well be filled tomorrow. You have to support Kubernetes, integrate with Kubernetes and play in the cloud-native world that Kubernetes dominates, but you probably shouldn't bet on making much money from Kubernetes itself. (Disclaimer: But I've been wrong before.)
Google's move toward hybrid cloud?
On a somewhat related topic, I was pretty surprised this week to see Google acquire a cloud-migration startup called Velostrata. Google has been beating around the bush of hybrid-cloud management for a while, and this move seems like step toward actually doing it.
Sure, the bigger plan is almost certainly to use Velostrata's technology to get customers onto Google's cloud, but check out this language from the Google blog post on the acquisition:
customers can control and automate where their data lives at all times—either on-premises or in the cloud—in as little as a few clicks.
Velostrata's press release (which you can bet Google vetted) includes similar hybrid language, and goes into a little more detail on how the technology works.
This is a smart acquisition in terms of playing catch-up with AWS and Microsoft by helping enterprise customers actually get their workloads into the cloud, but also in terms of making some money on the stuff they decide to leave on-prem. Especially for the company that released Kubernetes, one of the big promises of which is enabling multi-cloud and hybrid architectures, there's no use denying that corporate data centers still exist.
Hybrid cloud is also a consideration in this joint effort by Google and Pivotal to win StubHub's business.