Containers offer undeniable business benefits, because they equally empower both the developer and operator, allowing each to focus on what matters to them. In addition, the container ecosystem has captured the imagination – and the financial investment – of the entire tech industry, almost as though the Docker, Kubernetes, and Mesos documentation includes a secret new volume in the Game of Thrones series. Ultimately something has to change because while the buzz and the mindshare might be inline with these stratospheric valuations, if word on the street is to be believed, the revenue has not matched the hype.

However, it’s clear that the market in which software containers like Docker, Kubernetes, and Mesos operate is shifting. I see several things changing in the container ecosystem in 2017 – and they’re probably not what you’re thinking.

The Battle Lines Are Drawn

Over the past few years, I’ve watched each container community grow, both in its technical underpinnings and its industry role. Because the tools were new and innovative, it was easy for experts to suggest one over another; each one had strengths in different areas, so if you needed large scale you might choose Mesos, and if something more structured was important, it became obvious that Kubernetes was a better choice.

And, because the community saw opportunities for the entire product area to grow, everyone “made nice” with each other. A rising tide raises all boats, and so on.

As I predicted back in June 2016, I don’t expect that to last. As the technologies mature, we expect all of the software to grow and converge on a core set of capabilities, even if each tool uses a different approach. As a result, the choice between container runtimes is likely to become more competitive. After all, these companies raised obscene amounts of money, but they aren’t selling enough to justify their valuations. They have to distance themselves from one another otherwise, users will either conclude, “they’re all the same” or simply remain confused and make no choice forward.

The techies might still stay friendly; we generally do. We’re going to start to the growth of “neutral” foundations and standards proposed like Cloud Native Computing Foundation. Posturing and politics already are starting to emerge. It’s going to look like people are singing kumbaya – but I think it’s important to remember that investment dollars and valuations will ultimately drive strategy.

This past year we have already seen some small things change, but winter is coming, and this year I think the war will begin. We will see how these providers shake out with their revenue model, product features, and other business practices. I’m not sure who’ll win – and that’s okay for us since with {code} we can support everyone. We don’t have a dog in this fight, given that our projects are positioned to work with any of these tools. Those people attempting to build production environments might not be so lucky. The lesson here is that we have to be diligent about building platforms with proper abstractions so parts of the stack can be swapped as the industry evolves.  We don’t want to look back a year to 18 months and find that we are again locked into technologies that aren’t flexible or easily swapped out for more powerful options.

rex-ray-crystal-ball

In 2017, DIY isn’t the only option anymore

It’s an age-old challenge in the computer industry (and perhaps in everything else): Should you build a system that’s customized for your own organization, or buy a premade solution that does an adequate job (but is far easier to deploy). We’ve seen this journey with open source in every way, where organizations have to decide whether to build something themselves or buy a prepackaged solution.

In their hearts, everyone wants to build things themselves. But 99% of the world is simply not resourced to do it themselves, no matter what technology we’re talking about. So efforts to build an evolved platform either end up with an off-the-shelf purchase or they languish and never see production. The market offers plenty of “buy” options but they leverage relatively mature virtualized technology, taking an IaaS-first approach. These offerings are becoming less effective for long-term use because there’s desire to shift toward a PaaS approach. Developers and business units do not want to focus on infrastructure. They care less about instances of virtual machines, and more about instances of applications. Companies of all sizes are yearning for this application-first model, though there are very few pre-built options. Therefore, the industry must evolve past simply offering “reference architectures” and “validated systems” and instead not only package the software, but also reinforce modern DevOps principles for IT departments that lack the resources or culture to build their own.

So I think the market is ready, primed even for a “buy-it” option built around containers and a DevOps ecosystem. I believe that 2017 is the year this will happen. The conditions are right. Companies supporting container runtimes will have huge pressure to increase sales and an appliance vastly simplifies their go-to-market efforts. Further, customers want to consume this software, and appliances fit much more closely to their consumption model so entrenched in their organization.

What Happens in the Cloud When it Rains?

The cloud has been burdened by buzzwords and hype ever since it became feasible to rely on Internet services rather than (or in addition to) existing data centers. I don’t think we’re over the hype just yet – you can still hear people shout, “Everything is moving to the cloud!” – but things are beginning to settle down as companies find out which cloud formula works best for them.

In 2017, I predict we’re going to start to see a slowdown in people actually being able to run in the cloud. People will step back and realize it isn’t necessarily easier or cheaper. Sometimes it is – but the assumption that the cloud is always better is unrealistic and shortsighted.

This is part of a longer trend in which organizations learn to better judge whether the best answer for their need is found in the cloud or on-premise. By now, companies that struggled to move core services to the cloud have discovered that such a transition is not necessarily cheaper, and it still requires resources to do it – resources that a lot of companies don’t have.

Instead, I expect to see more organizations adopt platforms that allow them to run some stuff on-premise and some in the cloud. They’ll consider their own answers to, “How can we consume services easily?” – with solutions that may be in the cloud, on-premise or with a hybrid approach.

We will see a time of reckoning, where the choice on where to run an application is based on proven principles, not hype. And that’s a welcome change.

Do you agree with my assessments? Tell me if I’m missing something.

  • Pingback: Migrating a Flocker Storage Driver to REX-Ray | {code}()

  • At Pivotal, what we are seeing from F200 customers is that they want to differentiate their business with custom software, not snowflake platforms and customized container orchestration science projects that drive zero business value to the company. Buy don’t Build seems to be the future.

  • Rich Kucera

    In 2017, Serverless (FaaS and BaaS) will supplant “cloud”, “microservices”, containerization, SaaS, IaaS and PaaS.

  • Manny RodriguezPerez

    Solid assessment Josh – We see lot of discussion around these exact topics and our guidance is in line. I agree that the pace will accelerate this year under the “I have to do something” mantra , however I feel people are better informed. The only other trend that I believe will accelerate in 2017 is the Enterprise adoption of DevOps and software dev practices to operations.