***BOGO*** Re: running optimize/analyze command

["Nagaraj S" <nagaraj.chk@xxxxxxxxx>]

Chandru,

How to overcome by partition, Can you explain with details/examples

mdm,


-----Original Message-----

From: chandru [*mailto:pradeep.chandru@xxxxxxxxxxxx*<pradeep.chandru@xxxxxxxxxxxx>]


Sent: Monday, May 26, 2008 2:40 PM

To: Alex Arul Lurthu; Ananda Kumar

Cc: Sebastian Mendel; MySQL General

Subject: Re: running optimize/analyze command

Hi anand,

To add to alex's views, I see that you do two major operations 1. Insert

* Use a temporary table that loads the data initially and then you

copy the same to the original table, which can prevent load on the

original table and shall be faster as well.

* you can disable the indexes and then insert the records into the

table. after the insert is over you can enable the indexes.

* try using extended inserts and increase the bulk_insert_buffer_size.

* In case you use load data infile use that with CONCURRENT

* Share the I/O load by having the data written across multiple discs.



2. delete

* Please try to use partition if you can migrate that to MySQL 5.1.x

and partition the tables based on the delete condition.

* Since you where mentioning that there is huge volume of data that

needs to be deleted, check that if the max_allowed_packet value is

set to 1GB, if the selected data volume is more than 1GB try to

split the delete query into smaller ones

Note: confirm that indexes are available on the where clause

Regards,

Pradeep Chandru