If you're getting 100 separate requests for joining a level, can that be batched into a single request to return the entire set of surfaces? Or at least batched into chunks as groups of surfaces come into rendering range?
So that you don't have to hit the token table and other supporting tables as often?
The way it was written, not really. Essentially, each object that uses this is the originator of the request, it's all highly dynamic, so we can't necessarily predict the names (necessarily).
Will put it on the trello though, because even though the files themselves are like 50k each, I don't know if you can queue up a batch of them from Azure's blob service...