VMware-Heptio roadmap to refine enterprise Kubernetes

Heptio is now officially part of VMware, and company officials are ready to talk about their plans to meld Kubernetes management IP.

VMware closed its Heptio acquisition in December 2018. Craig McLuckie, one of the original co-creators of Kubernetes at Google and now VP of R&D at VMware, launched Heptio to guide enterprises through the many roadblocks to production container deployments. VMware, he said, will scale Heptio to reach more enterprises and blend Kubernetes with customers' vSphere infrastructures.

On the VMware side, Heptio brings upstream Kubernetes expertise, and forms a new entry-level Kubernetes distribution in the VMware product line, which will encompass Heptio's Gimbal ingress load balancer, Contour ingress controller and Ark data backup projects. VMware also will fold Gimbal, Contour and Ark into VMware Pivotal Container Service (PKS) and Cloud PKS service. Most importantly, Heptio will add training prowess that can bring enterprise IT pros up to speed on Kubernetes.

We caught up with McLuckie and Paul Fazzone, senior vice president and general manager at VMware, to talk through the details of the VMware-Heptio roadmap.

What stage of Kubernetes adoption are most VMware customers at right now? Is it still early?

Paul Fazzone

Paul Fazzone: Yes. Companies are learning how they get their first application deployed and operational. We're still 12 to 18 months out from mainstream adoption and operations in multisite, multi-application deployments. The Heptio field engineering team will be the seed for this inside VMware -- it's about helping educate customers on how to do Kubernetes the right way. It will be a year of going from enterprise newbies to midlevel and advanced-level knowledge of Kubernetes.

Craig McLuckie

Craig McLuckie: The hype is for a reason -- Kubernetes really does solve problems that haven't been solved before. But the movement and evolution of enterprise IT takes a lot longer than you might think. Even within an organization, you'll have very progressive groups that move quickly, and groups that take longer. As we start really digging in with organizations that deal with deep regulatory requirements and complex workloads, it's going to take time and there's an inordinate amount of work that needs to be done to realize that potential.

VMware is not alone in trying to offer those things to the enterprise. Rancher talks about unified management for Kubernetes in multiple environments, and Red Hat talks about that 'last mile' enterprise use case with OpenShift. The Cloud Native Computing Foundation says there are 75 vendors with some form of Kubernetes management.* How is VMware going to differentiate itself in a crowded market?

McLuckie: What motivated me to join VMware specifically relates to this problem. We were out there as a startup working with some of the world's biggest enterprises. It's fascinating to watch how fast their computational fleet has become fragmented. It's easy to get the first 80% of work done to get something working, but the last 20% tends to dominate the level of effort required to solve for real-world problems. Getting Kubernetes working in those [brownfield] environments is challenging.

What details can you share about the VMware-Heptio roadmap for 2019?

We're now looking at a collection of tools, SaaS services that will tie the various [Kubernetes] operating models under a common framework.

Fazzone: In the first quarter of our fiscal year, which starts in February, [we'll have] a range of options. What Heptio offered as its HKS product will anchor the portfolio as a base layer upstream-aligned Kubernetes distribution that we bring to customers with support. In the middle of the portfolio you'll see us offer what we now refer to as Enterprise PKS, which is packaged for customers who want to offer [Kubernetes] to their development teams while IT manages it for them. It won't allow as much flexibility or integration with upstream projects as Heptio, but it'll be structured for customers who want to get down to operating and don't want to spend time building and doing integration. We also introduced a product in beta called Cloud PKS last year for customers who want to offer Kubernetes clusters to their development teams, and don't want responsibility for the infrastructure underneath.

Eventually, customers will want Kubernetes-specific management and operational tools that can tie these disparate environments together. We're now looking at a collection of tools, SaaS services that will tie together the various operating models under a common framework. Heptio has also produced a number of projects -- Ark, Gimbal, Contour -- born out of identifying problem areas that were standing in the way of being able to operationalize Kubernetes in enterprise accounts. You can expect us to make those available across the entire portfolio.

How will Gimbal and Contour integrate with NSX within PKS?

Fazzone: We introduced a project that will go into beta in February called NSX Service Mesh to deliver higher-level networking and visibility services in and alongside NSX-T. The same core R&D team is responsible for NSX-T, as well as the NSX Service Mesh project. One thing we are looking at right now is how we take advantage of some of the capabilities of a service mesh architecture to help our customers with networking and security needs in Kubernetes clusters. There are unanswered questions there, but Craig [McLuckie] and Joe [Beda, formerly Heptio CTO and now principal engineer at VMware] are working with our engineering team to evaluate that. I expect us to have more specific answers in the coming months.

A lot of customers at KubeCon said their heads were turned by Istio. What's VMware and Heptio's outlook on that?

McLuckie: Istio is a pretty nascent technology. It benefited from the brand of Kubernetes, but it's almost a victim of Kubernetes' success in some ways. There's still work to be done in the open to converge it on a set of practical use cases for the enterprise. Istio is built around a lot of sensibilities that have to do with the Google view of the world, with large clusters that are relatively flat, but [enterprise IT] is a far more complicated and fragmented world. My advice to our customers is that it's something they should understand and think about, but it's going to take a while to bake. I would be hesitant to jump in with two feet, but I'm optimistic about the future of the technology.

*Dan Kohn, executive director of the CNCF, provided this vendor estimate during an interview this month with the author.