2017: The year of serverless?

As we embark upon a new calendar year, the blogs are full of navel-gazing looks back at the year just past and crystal ball perspectives on what is yet to come.

While I don’t go much for those sort of articles, having spent a few weeks recently in Israel and coincidentally talking with a bunch of people about serverless computing, it seemed a timely moment to reflect a little on something that humans suggest is the biggest thing since the advent of virtualization.

Serverless, for those unaccustomed to the term, is also known as event-driven infrastructure. Essentially it differs fundamentally from previous approaches towards infrastructure (physical and virtual servers and containerization) in that, rather than thinking about infrastructure as a series of increasingly compact units of computation, serverless thinks about inputs and actions. Serverless architectures, rather than articulating an “I want to achieve an outcome hence I need to create a unit of computation” instead come from a paradigm of “this input occur and therefore this action will be executed.” Serverless is therefore entirely focused on an output rather than the means to that output.

It is for this reason that many people have suggested, with -- it must be said -- a degree of hyperbole, that serverless completely changes the game.

I was reflecting upon this when talking to a venture capital firm that was looking for some advice about a potential investment in the serverless space. Said VC firm was concerned that serverless was a very early movement and that there was no clarity as to just how quickly it would be adopted by enterprises. And, since venture capital firms are all about addressable markets and market demand, this was a stumbling block to a potential investment.

I understood their reticence, but at the risk of crystal ball gazing, it strikes me that despite no real empirical evidence of uptake as yet, serverless is a journey that all investors need to be embarking upon.

Coincidentally, at about the same time as I was having these conversations, I got an email from Sirish Raghuram, CEO of Platform 9, a company offering services around open source projects such as OpenStack and Kubernetes. Raghuram wanted to give me his take on this year’s trends; in particular, he sees 2017 being the year that serverless starts to gain momentum. Per Raghuram:

"We will see serverless technologies gain real traction in 2017. Adoption will be led by the same, forward-looking DevOps communities that have championed container technologies, like the Kubernetes community did in 2016."

Which is, of course, the same sort of language that we have been hearing every January for the past decade or so heralding the coming year as the breakout year for some new approach (cloud, SDN, Docker, Kubernetes, serverless).

But the difference with serverless is that it isn’t simply a new way to package old stuff. It is rather a fundamentally new way of looking at the problems around the creation and maintenance of applications. Raghuram suggested that the rising tide of bots will be the vector by which serverless accelerates its market adoption. As he pointed out, a lot of DevOps automation involves writing bots that integrate with various systems using webhooks. A serverless approach makes this incredibly natural and easy.

This early awareness of serverless via DevOps automation will begin to sow a seed about broader use cases. And since many organizations have also been dabbling with containers, the benefits serverless brings to a container construct -- by way of easier and faster container management -- without having to understand a contained orchestration system, will prove attractive.

All of this leads me to suggest that the rise of serverless will be more rapid than we may otherwise think. It also leads me to question the 2017 prognosis for a number of companies and initiatives, in particular Docker, who have arguably failed to leverage the early attention they brought to the container world and are now at risk of being swamped by dual trends -- broader open source container management solutions such as Kubernetes, and a “forget the servers” attitude that serverless shops will have.

There are a lot of emergent trends going on simultaneously here -- containerization, serverless, bots, agility demands and the like -- and my sense is that these trends will all increase the adoption rate for serverless approaches. Which brings me back, full circle, to the conversation I had with the VC firm about investing in a serverless startup.

Yes, it is indeed early, and serverless is a technology that is largely unproven in the wild. But is also, I suspect, something that is going to get very big, very fast and, for that reason, something you should be looking at seriously.

This article is published as part of the IDG Contributor Network.


Computer World
http://computerworld.com/article/3155382/infrastructure-management/2017-the-year-of-serverless.html