Archives of the TeradataForum
Message Posted: Tue, 07 Feb 2001 @ 03:58:53 GMT
Other factors that you may want to look at are query load, how fast do you need to load the rows, and as James mentioned is table locking a factor.
Other options if you really need to speed it up, you can bundle your rows into groups and use multi-insert macros to do the inserts. You just define all the columns for all the rows in the group in the USING statement then put multiple inserts in the macro. For example we have a table with 700 byte rows, we can insert up to 3 rows at a time with a single execute of a macro. The more inserts in the macro the better the performance.
This method works very well for a constant inserting situation and provides high speed inserts with a minimum of resource usage. We have demonstrated rates exceeding 3000 inserts per second using this method on a 6 node 5200. At a more reasonable rate of about 500 inserts per second we saw less than a 10% degradation in query times.
Another option is fastloading into another table and then performing an INSERT SELECT into the target table. Under V2R3 and above the performance of INSERT SELECTS is very good.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|