List:General Discussion« Previous MessageNext Message »
From:Britske Date:January 16 2008 4:40pm
Subject:increase performance with inserts/updates in partitioned table when
knowing which partition to update?
View as plain text  
Hi, 

I have a partitioned table using hash. For the example lets say I have N
partitions.  Selects perform well due to the partitioning. However I noticed
that inserts and updates slow down considerably when the number of
partitions goes up. (I think because if needs to inspect / open all
partitions of the table)

Since hashing uses a modulo function I figured that if I inserted rows which
have the same modulo N in batches each batch of rows is inserted in exactly
one partition (correct?) 
I figured this would speed up inserts/ updates since only one partition
needs to be inspected for every batch. 

However, some measurements show me that this doesn't matter at all. 

Could anyone tell me what's incorrect in my thinking, or what I am missing, 

Thanks in advance,
Geert-Jan

BTW: I already asked this question, but wasn't sure the header asked the
correct question. 
-- 
View this message in context:
http://www.nabble.com/increase-performance-with-inserts-updates-in-partitioned-table-when-knowing-which-partition-to-update--tp14885273p14885273.html
Sent from the MySQL - General mailing list archive at Nabble.com.

Thread
increase performance with inserts/updates in partitioned table whenknowing which partition to update?Britske16 Jan