Arvados

Users would like to know the cumulative storage usage for a workflow. Should report separate values for inputs, logs, intermediates, and outputs.

Don't need to do full block de-duplication but it would be nice if it counted each PDH only once.

I'm envisioning that at the conclusion of a container, a query on the collections used by the container computes this number. The number is set on the cumulative_storage field which can be displayed in workbench and used by reporting tools.