AWS acquiescence with Kubernetes could prove a defining moment in evolution of cloud computing, paradoxically making AWS less differentiated over time
For the past decade, Amazon Web Services has been on a seemingly unending winning streak, launching innovative service after service with breathtaking velocity. This hot streak has had massive follow-on effects on the world we live in: cloud computing, in combination with ubiquitous use of smartphones created a perfect storm for a wave of innovation that has changed consumer experiences across every industry. All this while, AWS has appeared near invincible with an ever increasing range of services, market share and profits.
This week, at AWS’ annual re:Invent conference, that streak may be coming to a critical inflection point. Google’s launch of Kubernetes in 2014 is increasingly proving to be a stroke of genius, quickly passing AWS to take pole position as the platform for developers and operations teams looking to build modern cloud-enabled applications. At the same time, AWS’ own proprietary container service, ECS, has paled in comparison: Kubernetes has indisputably the best feature set, broadest industry adoption, and promise of portability across clouds and data-centers. Enterprise customers that have realized the lock-in intrinsic to AWS’ proprietary platform have increasingly turned to Kubernetes as a way to build cloud enabled applications that can run just as well on other clouds, or even good old on-premises infrastructure.
This has created a vexing conundrum for AWS: sit out the Kubernetes phenomenon, and risk losing workloads to other clouds and startups that preach the gospel? Or, throw in the towel and support Kubernetes, knowing fully well this will mean supporting the very framework that threatens to commoditize AWS by leveling the playing field across clouds and, triggering a catastrophic (for AWS) race to the bottom?
Without being in AWS inside circles, it is impossible to know for sure which way AWS will go. However, it seems likely that Kubernetes’ momentum has gotten to a point where it is table stakes for enterprises evaluating clouds; meaning AWS has no choice but to acquiesce and launch a Kubernetes-as-a-service solution, potentially at this year’s re:Invent. If this were to happen, it will likely prove to be a defining moment in the evolution of the cloud, with knock-on effects that could be rather surprising.
For one, supporting Kubernetes could mean that for the first time in the history of cloud computing, there is a single abstraction layer to deploy and operate cloud-enabled applications across any cloud. This means that enterprise customers gain leverage versus the big cloud providers. Every time an enterprise is concerned about escalating costs from their cloud provider, they could play off the big vendors and use Kubernetes to seamlessly divert workloads to the lowest cost provider. The implication: public clouds as a business could become a lot less differentiated and lose their highly profitable margins.
Second, not only does Kubernetes open up clouds to being cross-shopped more easily, it could also dull the value of other high-value application services these clouds provide. Consider, for instance, a popular AWS service like Lambda, for serverless computing. There are a range of open-source alternatives, such as Fission, which provide the same value proposition, while being open-source and running on any Kubernetes cluster. This is just an example of how there could be cascading shifts in value from cloud providers’ own native services to the range of application services that are available within the Kubernetes ecosystem.
Third, not only does this standardization reduce the differentiated value proposition of cloud providers, it could also reduce the perceived gulf between on-premises data-centers and public cloud providers. The many agile, well-funded venture startups operating in the container orchestration space will (partly out of necessity) focus on providing a great Kubernetes experience for enterprise environments running on-premises, in co-located datacenters and across hybrid environments. The result, paradoxically, is that private clouds could be easier to run, likely offering greater control and economics, while providing the same Kubernetes-based ecosystem of services that enterprises want to use in the public cloud. In turn, this will reduce the gulf between today’s public clouds and private clouds.
It is worth contemplating that impact will not be limited to AWS. Google’s cloud and Microsoft’s Azure could also feel some of the pressure from the commodification. However, two factors play in favor of these clouds: (1) they have historically been far behind AWS on market share, and have been looking for ways to level the playing field and win more workloads from AWS; and (2) both Google and Microsoft seem more open to investing in hybrid cloud technologies that could further de-position AWS as the leading enterprise cloud.
Whether all of this comes true or not, we’ll find out in a few weeks time if AWS’ falls in line with the rest of the industry on Kubernetes. If you’re an enterprise CIO reading this and wondering about the viability of private clouds and/or lock-in with public clouds, call your CFO: good times are coming.