Putting the server back in serverless

AWS releases new services by the boatload, and this year’s re:Invent was no exception. As their pace of innovation accelerates, one struggles to keep up with new features and rising services. But one emerging trend gets bigger every year: serverless.

Lambda computing since 2014 has been revolutionizing how new cloud applications are being operated and built, while enabling small teams of three engineers to scale to hundreds of millions of users. Lambda is permeating all layers of cloud computing, from traditional web servers to edge networking.

Does that mean servers are going away? If that were the case, it should be called server-least computing. Developers and applications can focus more on solving business problems and adding value, and less on “undifferentiated heavy lifting”, from scaling, compute, networking, and even databases.

Serverless databases?! A new AWS product currently in preview, even your database won’t require a server… or will it? The architecture diagram in the release reveals a “warm pool of DB capacity.” Yes, servers are still running your queries, but are also shared amongst multiple databases.

Abstracting the storage from the compute layer will do to databases what virtual machines did for bare metal hardware: higher density workloads, increased efficiency, and cost reduction in running you database.

Less servers will be used to run current workloads, but we should call this cloud-first database what it is: standby computing, reserve computing, or even “less server” computing.