Kubernetes is so hard—but worth the pain

2019 will be a challenging year for some enterprises as they try to turn Kubernetes hype into production reality

Kubernetes is so hard—but worth the pain
Thinkstock

It’s not even Christmas yet but the Kubernetes messiah has already been born. As a recent Heptio survey found, 60 percent of enterprises surveyed are running Kubernetes—and more than half of those are doing so in production. The Kubernetes training wheels, in other words, are off.

Which is saying a lot because, as Capital One’s cloud guru Bernard Golden has stated, “While Kubernetes-based applications may be easy to run, Kubernetes itself is no picnic to operate.” It’s not simply the learning curve associated with moving to containers and microservices-based architectures, but the newness of Kubernetes itself. That, coupled with a somewhat chaotic market for Kubernetes expertise, will make 2019 a challenging year for some enterprises as they try to turn Kubernetes hype into production reality.

Bringing order from chaos to Kubernetes

Kubernetes is plagued (blessed?) by “true goat rodeo dynamics,” as one industry observer put it to me. “From the old guard like IBM and Oracle offering Kubernetes products or services, and the ten-plus startups, and guys like Rackspace, and recently NetApp buying Stackpoint, Nutanix launching a service this week, Kubernetes is a crazy-fest.” There are, in case you were wondering, 61 CNCF-certified Kubernetes service providers. Granted, not all of them offer their own version of Kubernetes. Some, as IBM’s Jeffrey Borek correctly pointed out, are “incorporating Kubernetes into their platforms for performance and interoperability.”

That’s good.

And choice is also good, as CNCF executive Chris Aniszczyk highlighted to me (“Isn't it great that users have so much choice in support?”). But choice is only good up to a point. Too much choice ends up being as bad, if not worse, as no choice.

Everyone knows Kubernetes is a big deal, and everyone wants to pile on. But for enterprises trying to make sense of the vendor landscape, it’s frothy. No wonder that 46 percent of those Heptio survey respondents noted the difficulty in “aligning internal teams around common technologies” as their second-biggest challenge.

Which is not the same as saying it’s bad. It’s simply early economics: VCs are spraying cash at any startup with a Kubernetes angle, while lumbering legacy vendors are also trying to find relevance by hitching their wagons to the Kubernetes Kool train.

Why Kubernetes is so hard—but worth the pain

Back to Golden. As I quoted, “While Kubernetes-based applications may be easy to run, Kubernetes itself is no picnic to operate.” Even if you get everyone in your organization on the same page, using the same vendors and products, you still have to deal with the relative immaturity of Kubernetes. In Golden’s words: “It is an early stage product that is serving as the foundation of an ecosystem of supplementary products (such as Istio), all of which must be installed, configured, and managed correctly to make applications run properly.”

That’s not impossible. It’s just … hard.

But worth it. Sixty-three percent of enterprises running Kubernetes in production are “immediately using their resources more efficiently,” according to the Heptio survey, with 58 percent saying it has helped them to shorten development cycles. Those are goals worth enduring a bit of pain to reach.

So what will most enterprises do to attain this nirvana state? Golden said, “Expect to see a further enthusiasm for consuming Kubernetes from one of the big cloud providers [Amazon Web Services]; it specializes in running complex infrastructure at scale, and most IT organizations will conclude they should let the specialists take over.”

In the old days, new technology like Kubernetes would be digested by the legacy vendors and sold as their latest dish of magic beans. Today, however, enterprises don’t need to purchase the complexity that comes with the legacy vendors and will default to the cloud vendors.

Which means 2019 will be a banner year for the Big Three public cloud—AWS, Microsoft Azure, and Google Cloud Platform—driven in part by enterprises’ need to tame Kubernetes. The one dark horse in this is Red Hat/IBM, which takes on the burden of simplifying Kubernetes with its OpenShift platform, and appeals to more conservative companies. If Red Hat can break through the IBM clutter to energize its field, we could be looking at a Big Four in 2019, though with IBM/Red Hat playing a firmly all-hybrid, all-the-time game.

Copyright © 2018 IDG Communications, Inc.

How to choose a low-code development platform