There's little reason we don't have quotas now. However, at since tools with lots of usage would come to a screeching halt if a default quota of small size were universally applied, we should carefully set initial quotas based on the current size of a tool user's space consumption.

This implies we need:

• A process for handling filesystem quota requests with documentation
• An initialization script that retrofits quotas into the existing filesystem
• Monitoring for the subscription rate of our quotas so we know how deeply over-subscribed we are. Otherwise approval for expansion requests is just guessing.

This is a new process and major service modification, so this appears to be a proposal.