Hacker News Reader

Comment by twic | Hacker News Reader

I remember being in a meeting with a client, with a Pivotal salesman alongside me, when the question of PCF's overhead came up. We managed to work out that a PCF installation required about 10 separate VMs before you even got to the ones that would run applications.

View on HN

The point being that those VM can be distributed across multiple hosts for scalability and reliability. I'm sure that if you wanted to really package it all up into a single VM, you could... but that wouldn't make much sense when you're building an onpremis deployment cloud.

No, the ten VMs are not there for scalability or reliability, they're there because there are ten teams building the essential components of Cloud Foundry, and the BOSH model is that each component gets its own VM.

You could squeeze them onto one VM, but that would require that CF teams cooperate on manifests. Inconceivable!

Maybe you could fathom that the BOSH model is designed the way it is, for scalability and reliability purposes.

The CF team is a bunch of Pivots... they cooperate quite well and probably better than most other teams, especially since they rotate across projects fairly often. Disclosure, I worked on CF in the early days, on a number of teams.

PCF is for those working at scale, and doesn't suit trivial deployments well.

Someone really should have told the salesman that, then - they were trying to sell the client on the idea that they could have a separate PCF foundation for each project.

> they were trying to sell the client on the idea that they could have a separate PCF foundation for each project.

Cloud Foundry (PAS) is not designed to be used this way -- it had multitenancy designed in from the beginning.

For PKS the idea of cluster-per-project makes more sense, given that Kubernetes doesn't reaaaally have multitenancy in the same sense.

Agreed - that is not how the platform should be used.

Reply on news.ycombinator.com