The future is not Borg

The New Stack has an interesting write-up about Google's long-secret Borg system. I can't say anything specific about this and I haven't read the paper.

What I will say is when I first arrived at Google in 2005 I felt like I was stepping into the future. Tools like Docker, CoreOS, and Mesos are 10 years behind what Borg provided long ago, according to The New Stack's write-up. Following that delayed timeline, I wonder how long it will be before people realize that all of this server orchestration business is a waste of time?

Ultimately, what you really want is to never think about systems like Borg that schedule processes to run on machines. That's the wrong level of abstraction. You want something like App Engine, vintage 2008 platform as a service, where you run a single command to deploy your system to production with zero configuration.

Kubernetes is interesting to watch, but I worry that it suffers from requiring too much configuration (see this way-too-long "guestbook example" for what I mean). Amazon's Container Service or Google's Container Engine may make such tools more approachable, but it's still very early days.

I believe systems like Borg are necessary infrastructure, but they should be yet another component you take for granted (like your kernel, a disk driver, x86 instructions, etc).