In article <[EMAIL PROTECTED]>, "Andrew Hill" <[EMAIL PROTECTED]> writes:
> Hi, > Okay, cool. A couple of things: > 1) We *think* that our bottleneck is RAM and the calculation of indexes > when inserting into the table (about 500 million rows). There's > certainly plenty of CPU and disk I/O left in the machine. Hopefully, > with smaller tables, the inserts will happen faster as well, and getting > timestamps from MySQL won't be an issue. > 2) We're lucky with the report generation, in that everything can be > summarised into stats tables reasonably easily :-) > So, I guess that I'm more concerned with ensuring that the times are > correct, in case of the clocks on the clients going crazy. That's why I > was hoping there was a way of making MySQL select the table to insert > into based on the time... You could try allowing some inaccuracy: an entry might end up in the correct table, in the table "before" it, or in the table "after" it. This would require rewriting SELECTs with UNIONs, but because the UNIONs would generate only a few rows (if any), the performance should not suffer much. -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
