First things first

There's an increasingly common storyline told by industry pundits right now that serverless computing is going to rip the carpet out from under containers before they've even reached critical mass. The corollary effect of this is that Amazon Web Services will continue its market dominance because it was so early with its Lambda service, has seen Lambda adoption grow fast, and is pushing serverless very hard.

(I know, I just linked to two posts from the same author, Matt Asay, but I promise this is a broader conversation. Also, it's somewhat ironic that people are starting to come back around on AWS, after a period where it was in vogue to question its continued dominance (a la Apple in smartphones) in part because AWS did not have any real open source and/or Kubernetes strategy.)

The argument about developers wanting the easiest and fastest unit of compute -- a lambda or a function -- makes perfect sense, but I think it ignores (at least) a couple of key things. One is that open source really does matter, especially at the infrastructure or foundational level, and it's not clear a proprietary service on a proprietary platform will be what wins in the long run. Even if serverless does come to dominate, there's no guarantee Lambda will be the industry standard (Amazon EC2 is the industry standard for pure IaaS, but really it's Linux that dominates at layer most people care about).

The other thing is that while developers get most of the love, devops is a portmanteau comprised of two words. And the second one, operations, is still very important. I think that what Kubernetes, containers and the whole microservices / cloud-native movement have demonstrated is that at companies of any meaningful size, operational concerns around security, reliability, visibility, manageability, portability, etc, etc, will continue to drive IT decision-making. (You can get a sense of this in my August podcast interview with GitHub SRE Jesse Newland and also in his blog post explaining the company's move to replatform certain piece on Kubernetes.)

Serverless can still win out as the preferred method and architecture for developers, but ops teams and engineering execs are going to decide how serverless is consumed. For folks who prefer to just use a ready-made service, all signs point to AWS Lambda as the market leader now and going forward. But for folks who want to maintain some level of control at the lower layers, I think some sort of serverless platform built atop Kubernetes is going to be the preferred option. I suspect this is where serverless efforts within the Cloud Native Computing Foundation will ultimately be focused, and indeed is the whole idea behind the Kubeless project.

Among the openness and control set, Google seems to be in a better position to own future workloads. Just look at what it announced this week, with the open source Asylo framework for developing containerized applications that run in hardware-based trusted execution environments, and with a point release of Kubeflow, a an open source project dedicated to running TensorFlow deep learning jobs on top of Kubernetes.

On their own, neither of these releases constitutes an industry-changing effort, but look at the big picture: Google is building for a world where Kubernetes is the OS, as it were, and it's going to make sure Kubernetes checks all the boxes that users care about. Open source, controllable, composable, secure, artificial intelligence and, yes, even serverless. Whatever the love triangle situation is among Netflix, Amazon and Google, I think it's openness and devops/AI prowess that brought Netflix and Google together in the first place.

Of course, who ultimately reigns as the cloud market leader is subject to any number of other factors, including the evolving strategies of companies like AWS, Google and even Microsoft. (And, internationally, Alibaba.) But right now, if I had to bet on a handful of AWS services against the community currently rallying and growing behind Kubernetes, I'd go with community.

On somewhat related note, Google scored Twitter as a cloud customer this week, and AWS scored Oath (nee Yahoo). Scroll down to the Cloud and Infrastructure section for my thoughts on that news and links.

Read and share this issue online here.

Sponsor: MongoDB

AI and machine learning










Sponsor: Replicated

Cloud and infrastructure









Data and analytics