On Wednesday 20 February 2002 02:06, Anvar Hussain K.M. wrote: > Hello, > > I don't feel the the table sizet will be a problem. But how would the > database function with such a huge number of queries per second. > > If I read right, you will be taking a maximum (presently) of 42 channals > each taking taking data at the rate of 360/sec. That would be 42x360 = > 15120 queries to the database per second. Can MySql withstand this?
My concern wouldn't be so much "can MySQL handle 15120 queries per second" but can your disk drives handle it... Even with some pretty high-end hardware thats a tall order. Consider, a top line SCSI drive is 10k RPM, that means that at the very best the heads get over each sector on the disk 10k times per second, so that is the absolute theoretical maximum rate at which reads and writes can be issued to the drive (since drives have to spin around once per I/O operation no matter what size it is). Given that many operations span multiple cylinders that is just the optimum rate of the hardware, and doesn't account for seek time delays, which are orders of magnitude higher than that. Of course the controller and the OS, and MySQL can all help by coagulating various operations into lesser numbers of larger operations, but fundamentally the limit is still there. Disk stripping and to a somewhat lesser extent, other forms of disk load balancing will also alleviate the bottleneck, but you are still talking about a very high level of write frequency. I'd be loathe to claim that any particular system would handle that load without testing that hypothesis. Its certainly doable, but might take a bit more hardware than one would think at first blush. > > One solution is to insert multiple rows in a single query. Possibly > inserting only after a monitor has taken 100 samples (an optimal value can > be found by trial). Now the number of queries would be reduced to 3.6 x 42 > = 151. Something manageable for the database engine. It would do good to > keep number of indeces to the minimum required and simple. > > Anvar. > > At 07:29 PM 19/02/2002 -0800, you wrote: > On the largest plants we would expect to monitor, we would have about 42 > separate analog channels (tables) taking data at the 360 or more times per > second. Each of these tables would have two fields - a record number > (index) autoincrementing, and a numeric (probably a smallint in most cases) > which are change, and a couple of static identifier fields, so each record > would be on the order of 100 bytes or less. > > Michael Mitchell, > Principal, > DGRI Systems > > > --------------------------------------------------------------------- > 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 > > > > --------------------------------------------------------------------- > 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 --------------------------------------------------------------------- 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
