leungchopan - Fotolia
Kubernetes deployments have become a more common IT practice over the last three years, but the technology is still too difficult to manage, according to one of the industry's foremost Kubernetes experts.
Rancher co-founder and CEO Sheng Liang based his business on addressing this problem, and announced he would sell it to Linux vendor SUSE in July. This week, the acquisition officially closed, leaving Liang free to discuss his strategy for solving the Kubernetes deployment challenges that still face SUSE's base of large, traditional enterprise customers.
In Liang's view, being part of a larger, well-known enterprise company will give him a stronger foundation to simplify Kubernetes deployments in those environments. Meanwhile, SUSE's product line will be brought up to speed with cutting-edge container orchestration techniques through Rancher's Kubernetes expertise.
The acquisition has closed. What does this mean for SUSE and Rancher customers in the coming year?
Sheng Liang: As a small company, Rancher was good at building innovative technology, and pushing the market forward. With SUSE, the alignment around open source is, I would say, probably the strongest amongst all the other possible places that Rancher could end up. And SUSE is much bigger. That allows us to provide customers with much better backing as they rely on the Rancher platform.
In the last few years, I've been in quite a few calls, where the VP of infrastructure or the CIO said to me, 'I just want to make sure you're going to be able to give us good support, and you're going be around not just this year, but five or 10 years from now.' With Rancher being part of SUSE, we put all those concerns to rest.
Rancher brings Kubernetes deployment expertise to the table here. What do SUSE's engineers bring to Rancher that might show up in its products?
Liang: SUSE has built some unique expertise and differentiation around adaptable Linux, not just for enterprise data center and cloud, but they're very strong in edge and embedded systems, which aligns well with the recent Rancher focus around Kubernetes on the edge. You can imagine the Linux system, say, that's running inside a car may not be exactly the same as the Linux system that runs in the data center. Customers probably need a special kernel build, they may need some special drivers or configuration, but it's dangerous to do it yourself -- they want a partner like SUSE to stand behind it.
Liang: Rancher is agnostic to the underlying Linux distro, the Kubernetes distro or Kubernetes service, and that's usually one of the first questions we got, once we announced this acquisition with SUSE - 'Does that mean you're going to stop supporting Red Hat Enterprise Linux?' And that's not the case -- all of that can continue, because SUSE is fairly unique. If you're not careful, open source vendors can still lock you in to a particular set of integrated technologies via certification. [They'll say] 'You can do that, but we will just never support you, if you run our software on something we don't bless.' That is just not how SUSE does its business.
What's the upshot of the combination of that adaptable Linux for edge and embedded apps and Rancher k3s? What features will joint development make possible now?
Liang: To be honest, I personally have not put a tremendous amount of thinking into a proprietary tie-in amongst the different parts of the SUSE solution stack. I think there might be opportunities, but what SUSE would be more likely do is to upstream that technology, either into Linux or into Kubernetes.
At the same time, SUSE is not a nonprofit, right? So, there's going to be some way to extract business value from this acquisition.
Sheng LiangPresident of engineering and innovation, SUSE
Liang: SUSE has been tremendously successful -- it just had a fantastic quarter. And that's without Rancher. The reality is if we build a product that adds value, and then we build a support organization around it to make sure that customers trust us with their mission-critical workloads, that enterprise subscription model works. SUSE has been around 20 years. Rancher's been perfecting that business model over more than six years. And we're a true believer in it.
The other thing is, earlier this year we launched our venture into Hosted Rancher, essentially bringing us closer to a SaaS, which would be a different business model that could also be used to monetize open source software. SUSE's business on public cloud providers has been one of their huge growth drivers. Those are some of the areas I think you could also see us starting to innovate. We happen to believe support subscriptions are a great model to innovate and to monetize open source, and we don't see the ceiling of that yet, but that doesn't mean we're not open to either a proprietary add-on, potentially, if it makes sense, or to further develop these Hosted Rancher services into a true SaaS.
Meanwhile, Amazon announced EKS and ECS Anywhere, and they're going to open source their Kubernetes distro. Other cloud providers also sell hybrid, multicluster management, such as Google Anthos and Azure Arc. What will be SUSE's competitive approach there?
Liang: From our perspective, Kubernetes is a commodity, a standard for compute, almost like the TCP/IP protocol standard for networking. A lot of vendors, including Amazon, Google and us, we like to claim that we know Kubernetes better, we understand how to build a distro or a community service. But I think the upside of that is actually fairly limited. It does make a lot of sense for Amazon to keep pushing it, because they're fundamentally very good at infrastructure, and as Kubernetes becomes the standard, they want to be very good at supplying that infrastructure. But from our perspective, we are pragmatic about Rancher or SUSE's ability to be the distro provider.
The opportunity we saw was not being a supplier of Kubernetes distros, but given that all these consistent Kubernetes distros are out there, there's just a tremendous opportunity for enterprise IT organizations to finally simplify application deployment across all infrastructure. So, Rancher is actually not Kubernetes. We have a Kubernetes distro, but these days, it's all based on k3s, very much targeting edge use cases. If a customer goes to cloud, we'd say use EKS. If a big customer said, 'I also want exactly the same EKS distro as I have in Amazon on a bare-metal cluster,' we would welcome that too. In fact, we'll probably do a lot of integration to make sure people can easily do that through Rancher. What Rancher brings on top is managing all the Kubernetes clusters that are everywhere.
Kubernetes does seem to be a commodity. Even at KubeCon, most of the discussion was about things like service mesh, and observability. Rancher already integrates a lot of those things, but what is left to be done there for Rancher in the Kubernetes ecosystem?
Liang: In theory it is, but I can tell you now, it's still way too hard for people to consume Kubernetes. Service mesh probably made it even harder -- driving service mesh adoption has actually been a very tough hill even for Rancher to climb. We haven't gone very far there. It's not like the technology is not available, even mature. It's not even that some people wouldn't be benefiting from it. It's just we couldn't quite get a critical mass of all of our users to use it every day. To some degree, Kubernetes falls in that category, too -- there are certainly fans of Kubernetes, they get a lot of benefit out of Rancher. But if you look at the ubiquity of Kubernetes versus the ubiquity of Linux, there's a long way to go.