Arvados

h1. Federation use cases

I can view/edit any Arvados object I want, just based on its UUID or PDH -- no mucking about with credentials / environment variables to select a specific cluster.

I can do a search on multiple sites at once, and show the combined results.

I can see a web dashboard with all the stuff I’m working on (even though it’s spread over various clusters).

I can move my “home” account from one cluster to another.

I can run a container on a private site that accesses data from both private and public sites.

On a public site, I can share files/objects with a user who has never logged into this site (but does have an arvados account somewhere else).

I can create a project that contains projects/items from various other clusters (so I can share/unshare all that stuff as one logical unit).  All of the contained items appear on the “show project” page even though they are from different clusters.

I can share files/objects on one cluster with a group that is defined on a different cluster.

I can develop a workflow on private site A, and easily move it to private site B when it’s ready.

I can browse/select/use a public workflow (site A) on a private site (B), using some private data from B and the public data on site A, without actually copying the data from A to B.