IBM HashiCorp & HPC

On the face of things, the HashiCorp acquisition by IBM doesn’t seem particularly relevant to the high performance computing game. To see the importance of it, we need a brief history lesson and to take a look of some of IBM’s previous acquisitions.

The story starts in 2012 when they acquired Platform Computing, creators of the Symphony HPC scheduler. It was, and arguably still is, the most advanced commercially available HPC scheduler. Certainly, if you intended to share hardware across multiple competing HPC applications there was (is?) little else to compete with it.

Fast forward 7 years to 2019. Peak Kubernetes. Everyone wanted to run everything in a container under K8. And IBM acquired RedHat, makers of OpenShift. (If you didn’t know OpenShift is to K8 what RHEL was to CentOS… though that relationship is considerably more tarnished these days).

And now we have the final piece of the puzzle, Terraform by HashiCorp. The ability to spin up cloud resources across multiple providers with a unified language.

So, in theory at least, IBM have all of the bits they need to give us HPC, on any cloud on containers or VMs. The potential is enormous.

I could stop there, Perhaps I should. IBM certainly won’t thank me for the next bit.

I’m not holding my breath. 

IBM have owned RedHat for 5 years. Have we seen a revolution in how we run HPC with containers?

We’ve run multi cloud HPC. Terraform wasn’t the right choice to control the cloud resources. In its current form, I still don’t think it is.

So yes, there is massive potential here. I’m yet to be convinced that IBM will fulfill it to deliver what could be possible.

Feel free to tell me I’m wrong. It has been known to happen 😊