On 2002-01-30 10:08:12 -0500, Nick Vargish wrote: > On Wed, 30 Jan 2002, Marcus J. Ranum wrote: > > > Which nicely raises the next thing I wanted to ask: Does anyone have > > any statistics for how many records/second you can insert into a > > MySql database for a reasonable set of assumptions? I.e.: > > Is there a reason you are thinking about MySQL and not other RDBMS's? > > MySQL uses a table-level locking mechanism that will probably really > hurt performance if you're doing inserts from more than one client... MySQL is a poor performer for complicated insert operations. However, for logging it might be ideal, since you are appending to the end of the table and actually don't need any locking for the average case. MySQL is very fast indeed for query operations if you use the default MyISAM tables. Speed for other table types is much slower - on the order of PostgreSQL speed (i.e. slow as molasses, about 20x slower in my tests). Query speed would seem less important to me for logging, but if you want to do statistical work it might become more so. But our application is extremely query heavy (avg. 372 queries/sec vs. tens of thousands of updates/day) so I don't have any numbers for insert speed. :( We opted to use text files for our logging because after one day the log database would be much larger than our main database. MySQL does have the advantages of being cheap and easy. A sysadmin can easily install and administer a MySQL database, not necessarily true for, say, an Oracle setup. -- Shane Carpe Diem --------------------------------------------------------------------- To unsubscribe, e-mail: loganalysis-unsubscribeat_private For additional commands, e-mail: loganalysis-helpat_private
This archive was generated by hypermail 2b30 : Wed Jan 30 2002 - 08:37:08 PST