A table access method can decide itself if it wants to use the PG buffer cache, pgfdb does not so it's irrelevant in that case.
Also, how do the deployment semantics of FDB affect this. If I remember correctly, you typically run one FDB process per CPU on a machine. How do PG processes map to those
You would run the two separately, so PG processes would run and scale separately from FDB processes. Postgres spawns a new process for each connection so you would not want to map Postgres itself 1:1 with CPUs!
Comments
A table access method can decide itself if it wants to use the PG buffer cache, pgfdb does not so it's irrelevant in that case.
You would run the two separately, so PG processes would run and scale separately from FDB processes. Postgres spawns a new process for each connection so you would not want to map Postgres itself 1:1 with CPUs!