Archives of the TeradataForum

Message Posted: Tue, 29 May 2001 @ 18:27:26 GMT

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

The Teradata compression documentation is being completely rewritten for V2R4.1 to bring the documentation up to date and describe Teradata compression in a more positive light. Many of the limitations in the current manual are out of date and don't apply any more. The V2R4.1 documentation will dedicate an entire chapter to compression.

I have collected over a dozen testimonials on Teradata compression and many report a substantial savings in capacity and an improvement in scan query response time. I can't post the testimonials here because I haven't yet asked the customers for permission to disclose the specific information

The only potential negative I have heard so far is that for a large MLOAD going against a table with a NUSI on the compressed field(s), the MLOAD may take longer to complete. I'd be interested to learn if anyone else has experienced this and has some measured data or any other downside to compression.

So far, from what I've heard the benefits greatly outweigh any potential downside.

It should be noted that it appears that any additional CPU used by compression/expansion is more than compensated for by the savings in the CPU use for I/O. Anecdotally I have heard of 3-5% less CPU use for a fixed workload for a customer achieving about 33% compression savings.

Mark Morris
Teradata Architecture Group

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