Thanks, John. That does ring a bell, that it was maybe the number of database files, not the number of total files... And yes maybe (hopefully) it doesn't work that way in 9.5 (which the client isn't on yet)?
Karen In a message dated 5/30/2012 3:08:26 PM Central Daylight Time, [email protected] writes: > Karen, > > There was a thread in August 2010 – I recall John Minyo commenting on > this but I haven’t managed to find his post. What I found at the time was as > follows:- > > “....I recently had an issue with the amount of CPU time one particular > database was using. This was due to the number of RB? files I had in the > directory (all but one set being the database in use, the others being > renamed > superseded versions) – deleting these reduced CPU usage on a two CPU > machine from a continuous 50% (i.e. essentially all of one CPU) to a minimal > amount.” > > At the time I used procmon to see what was happening and saw that RBase > was ‘looking at’ all the RB? files in the directory even though they weren’ > t connected. While you were presumably referring to the number of files in > general, my experience then was that the number of database files had a > major effect on performance. I tried adding a few redundant *.RX? files to > a > directory yesterday with 9.5 but it did not happen then, so perhaps > something has changed. Maybe John Minyo can provide an update on how this > works > now. > > > > Regards, > > > > John Docherty > > >
