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 s

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!