Jackson Miller wrote: > On Saturday 16 August 2003 4:41, Hans van Harten wrote: >> Depending on your exact needs, consider brutal overwrites: >> LOAD DATA CONCURRENT INFILE '...' REPLACE INTO TABLE >> Replaces 10000 rows within 3.5 s into Innodb running MySQL-max-4.10 >> and WinXP on Celeron /400MHz/ for me. > Is there a need for CONCURRENT when using InnoDB? Isn't that the > default for InnoDB? True, CONCURRENT has no effect on InnoDB. I just grabbed a line of code and (made you) hit a tombstone: changed only recently to Innodb by 'default_table_type=innodb'. Everage response time dropped about 30% with bearly no effort.
> If not, does it lock the entire InnoDB table while the insert is running? Just to run a test, I repeated the 10000 records 100 times in a single file and imported that bulk in about 195s. During these 3 minutes multiple queries from another PC were answerred -in about 30 times normal response-. That proofs the absence of table locks, I'ld say. However, I fear the near-hog of the CPU is of an even more nacking concern. Makes me consider to do INSERT DELAYED INTO tabel (column) VALUES (1),(2), ...(n) and throttle a 1000 rows at a minute interval. What is your current rate of records/sec and CPU-load while using single row INSERT ? How does it effect response time to normal clients ?? HansH -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
