I've noticed the following strange behavior on Firebird 3 latest release, 64 bits, on Windows 2012, 24 processor server, 200 GB ram, magnetic HDDs (no SSDs):

1) After executing and commiting a massive delete on a table, any select or insert takes time to execute, as if FB was not reclaiming the deleted "space" on the table, and had to read each deleted record to filter and find the response.

2) While a big transaction is occurring (inserting into a table the results of a big select via an SP), another DB on the same server becomes unresponsive for about a minute. I thought each database had it's own "deamon / process / thread" assigned and the locks were exclusive to each DB, however it appears there is some interaction between the two DB performance other than just the shared OS/HDDs/Memory/Processors.

And a question for the experts would it be possible to install multiple instances of FB 3 on the same OS instance, to ensure a better level of independence between the DBs response time? This was possible on FB2.5 however I believe on FB 3 it did not make sense because FB3 assigns a processor per "request / connection", perhaps I got that wrong?