We have run into issues with giant recently where, for large workspaces, the performance suffers. So far, here are our ideas around the cause of this:

• When there is a job in progress, workspaces repeatedly poll the /workspace endpoint to fetch the full json forthe whole workspace. The size of this json response can be very large (28MB recently)
• We continue to poll even if the tab isn't in focus
• Where giant is struggling, we may scale down workers which will mean more jobs are in progress for longer, exacerbating this issue
• We'd like to check the performance impact of mixing in the Remote Ingest jobs into the workspace view - is this an expensive neo4j query?

Possible fixes

• only poll in focus tabs (could use https://developer.mozilla.org/en-US/docs/Web/API/Document/hasFocus or window.onfocus/window.onblur)
• poll less frequently for massive workspaces
• improve the performance of the neo4j query (sounds hard given that we are basically asking for 1000s of leaves of the workspace tree)