Docker and some serverless ambitions

Interesting…. I wasn’t at DockerCon last month in Europe (long story) but was vicariously attending watching the coverage from my two commentators of choice, Alex Williams at The Newstack and James Governor at Redmonk.

In his post-event review, Governor commented on one of the more interesting quasi-announcements at the event. Quoting from Docker CEO Steve Singh, Governor explained that he articulated the view that:

Docker is defining a comprehensive platform, a platform which can run any application on any platform. It needs to support any platform, packaged apps, Windows, Linux, mainframe, for portability and independence. It needs to allow for plug in, edge/IoT and be secure by default. We need to deliver a platform which is plug and play. We have to deliver serverless anywhere. We need to deliver an amazing experience. Serverless needs to be anywhere, which can’t be tied to any platform [italics by Governor]

Bear in mind that this keynote occurred during the same conference where Docker announced its capitulation with regards container orchestration: whereas it had historically tried to push the singular use of its own orchestration product, Docker Swarm, the company announced that it would provide a level playing field with support for both Swarm and everyone’s flavor of the day Kubernetes. So to have Docker move the conversation on not just the orchestration front, but also articulate an opinion about serverless, was telling.

Apparently nothing more was said at the show about serverless, but in follow up questions to Singh and Docker founder, Solomon Hykes, Governor was given answers that, as he understood it, suggested the Docker corporate view that:

the market needed to coalesce somewhat before Docker made it’s move. Premature packaging can be dangerous.

Looking at the serverless landscape

Some more exuberantly-minded commentators suggest that serverless may well be the technology that renders traditional virtualized infrastructural approaches obsolete. While it may be a little premature to suggest that, it is fair to say that there is some massive interest around the general serverless concept. Since AWS introduced Lambda, its serverless offering, a few years ago, many vendors have been racing to join the party.

Microsoft obviously has Azure Functions, Google has its own Functions product and there are some open source initiatives underway to offer a portable serverless offering – IBM initiated OpenWhisk and has seen some early backing from both Red Hat and Adobe. Then there is OpenFaaS which has been causing a bit of a stir within the CNCF ecosystem – ironically, when it comes to Docker’s moves, at least – the home of Kubernetes. Even Oracle, normally a company to dismiss early-stage technology trends, has come to the party with its own Fn offering.

Can Docker be meaningful in the serverless world?

Reflecting on Docker’s articulation of a possible serverless play, and the fact that they are somewhat late to the party, Governor pointed out that:

First mover advantage in tech can be somewhat overstated – just look at Facebook, for example, mopping up others’ ideas. Docker is in a position to embrace the technology that the New Kingmakers adopt.

I agree with Governor in his assertion that, simply being a fast (or even a not-so-fast) follower doesn’t preclude a vendor from being a powerhouse. Perhaps the better analog for me is Microsoft. Redmond was famously slow at jumping on the cloud computing bandwagon and even made several missteps at the start. That hasn’t however, impacted upon their ability to very rapidly pick up marketshare and secure a strong second-place position in the cloud infrastructure market.

MyPOV

For me, however, this isn’t a question as to whether Docker’s late entry into serverless will have an impact on whether they can execute on an opportunity or not. Rather, this is more a question of whether Docker’s long and somewhat convoluted path to where it is today, alongside the commercial realities of where it finds itself, will allow it to successfully go down this road.

Docker has used up a number of its lives already, and been either lucky or smart (depending on your perspective) to find ways out when things got tight. Whether serverless will allow them to do this in a world where container orchestration has largely resolved to Kubernetes remains to be seen. Hykes and co have a great ability to navigate twisted and torturous paths – it will be fascinating to see how they continue to do so.