Archives of the TeradataForum

Message Posted: Thu, 01 May 2003 @ 08:53:04 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Hi,

There was a presentation done at Partner 2002 by Anita Richards, called "A cross comparison of Data Load Strategies".

Basically a study was done comparing the different load strategies, where 5 million rows of data is required to be loaded daily into a table containing 3 years of data, with no secondary indexes.

The following tests were performed:

- Multiload
- TPUMP
- Multiload into a table with monthly partitions
- Fastload / Insert Select
- Fastload / Insert Select with monthly partitions

Three result graphs have been produced, showing the following:

- % of daily resource by load every x days
- % of daily resource by load every x hours
- % of daily resource by load every x minutes

The graph showed that TPUMP used a consistent 14% of daily resource.

It also showed that TPUMP resource usage was less than Multiload and FastLoad / Insert Select for any loads more frequent than daily. However when monthly partitions are introduced (V2R5), TPUMP resource usage was only less, if the frequency of the loads are less than 4 hourly.

I did wonder if TPUMP also benefitted from the monthly partitions, but this seemed to have been missed from the study.

If you email me off the forum, I will send you a copy of the presentation file.

Regards

Matthew Winter

Technical Architect
TPG IS, Design Authority
Teradata Certified Professional
Teradata Certified SQL Specialist

	<Prev	Next>		<<First	<Prev	Next>	Last>>