Re: Strange behaviour of the buffer pool

Looking at a larger instance on the same versions I can see the "resident" pool being 354MB while the "res-buff" being the same the BUFFERS setting: onstat -g mem|grep " R " res-buff R c00000001669d040 1280004096 2848 2 2 resident R c000000000491040 371245056 2848 2 2 onstat -c|grep BUFFBUFFERS 625000 # Maximum number of shared buffers PHYSBUFF 320 # Physical log buffer size (Kbytes) LOGBUFF 136 # Logical log buffer size (Kbytes) And as onstat -R and onstat -P shows that only the number of buffers specified by the BUFFERS setting are being used by the buffer pool, I assume that the rest is just used by some internal structures etc. But this is still quite a large overhead, 224 MB for the smaller instance and 354MB for the larger one. The SQL statement/dictionary/stored procedure caches are all default.