As our customer base increases, so are the usage patterns of our users. We saw an increase in very big data queries the last weeks, which our queuing system handled efficiently but not very fair-minded with respect to users that wanted to see their small queries fulfilled in a reasonable time frame. To accommodate all of our users, we had to tweak the queuing system of LOBSTER.
From now on we have a fixed amount of threads (currently three) that run in parallel associated with each user. Of course you can still enter all your queries at once, and three of them will always start immediately while all others get queued until the first one finishes.