Re: Reducing checkpoint times on cooked db

BART FISHER wrote: > The database is embedded in a product and > by default will use cooked files because of > configuration issues with setting up raw files. You could just create it with the cooked files, take an archive, change the cooked files to links to raw disk, and restore the archive. > I thought it will require additional writes to the > database because the dirty buffers will get written > by the cleaners more often and you will not get the > advantage of multiple writes to the buffer before it > is put onto disk. Yes, lower LRU_MAX/MIN will result in additional writes to the database; the point being that these are not all waiting for the checkpoint before they occur. The idea was that doing 100 writes spread out over 5 minutes was better than going 5 minutes with no writes, and then doing 75 all at once. Well that certainly is a big performance hit you're taking, but you didn't tell us how many LRU's, page cleaners, and AIO VP's you're using, and how many disks, dbspaces, and chunks you have. Maybe you just need more AIO VP's. June -- june_t@hotmail.com Lost in the wilds of Palo Alto, living on Peanut M&M's