On Mon, 14 Jan 2002, Benjamin Arai wrote: > Raids don't really help Table performance because in almost all cases the > bottlneck is caused by the drives access time. raiding drives doesn't > increase the access time therefore, you are most likely not going to see > and poerformance increases using a raid system unless you are change to > drives to ones with lower access times.
While I'm sure it's possible to "prove" this statement with a contrived benchmark, it's not at all true in the general case. A striped RAID array will drastically increase performance in real-world tests, where many simultaneous queries are taking place over "random" sections of the database. Think about it: access time is a function of how far (on average) the head will have to move to access the appropriate cylinder, and how fast the drive is spinning (i.e., how long you have to wait for the appropriate sector to appear under the head). If you have only one head per cylinder and you're accessing random cylinders, access time will be at its worst. If you have a head per cylinder, access time will be at its best. In a striped array, you're increasing the number of heads per cylinder, so you're decreasing the chance any given head will have to move, so average access time drops. james montebello --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
