 |
 |
Archives of the TeradataForum
Message Posted: Mon, 16 Jun 2003 @ 13:45:33 GMT
| Subj: | | Re: PPI's & hard-coded predicates |
| |
| From: | | Fuller, Joe |
Greetings all, happy Monday!
Actually, I would suggest V2R5 using the Partitioned Primary Index. It kinda creates a separate "table" for each date partition and is
extremely fast and functional. We benchmarked Call Detail Records (CDRs) at the benchmark center last year in the order of 80 Billion rows,
with a partitioned primary index of the Call date. Date range queries, including the BETWEEN clause, were the fastest I have seen. One of
the nice features of PPI is the ability to add new dates to a range, and drop the old dates. Dropping the old partitions effectively
deletes the data for those dates. So, if you have a table with 60 days rolling data maintenance is very easy.
Best Regards,
Joe Fuller
Sr Data Warehouse Consultant, DBA
Teradata Professional Services
Advanced Teradata Certified Professional
Teradata Certified Designer
| |
