Re: LRU Parameters

Also sprach "Scott Kolaya" <skolaya1@nycap.rr.com> : :Values of: : :LRU_MAX_DIRTY 3 :LRU_MIN_DIRTY 2 : :are commonly used for OLTP processing to decrease the length of a :checkpoint. How long are your checkpoints taking? They should be very :short with those values, but your write cache will suffer. Actually, the "write cache" as reported by onstat -p is a chimera. ALL writes, from a user perspective, are cached, i.e. any database update occurs against a page in the buffer. The write % cached is simply telling you the ratio of physical i/o to logical i/o. Since physical i/o is asynchronous to database update activity, as long as it is kept at a level that is not otherwise affecting system performance adversely, then who cares what the write cache rate is? What matters is checkpoint duration, as that has a direct affect on database activity. Dave (posting from home, speaking as a representative of nobody but himself) ********** "Everybody plays the fool, sometimes."