Re: Partitioning Time Based Tables in Oracle8

start_time/end_time optimisation is always a pain if you stick with simple tables and SQL.

In you case, you need to partition by one of the columns and create a local index on the other.

The column to partition by is the column which will usually result in the larger number of partitions being eliminated - it's your data, you have to make the choice.

If you partition on start_date and index on end_date you might want to consider a DESCENDING index (if you are on 8.1)..

--

Jonathan Lewis
Yet another Oracle-related web site: http://www.jlcomp.demon.co.uk

neilgreene_at_earthlink.net wrote in message <7sdctk$703$1_at_nnrp1.deja.com>...
>I am looking for a recommendation to partition a "time-based" table. A
>table where records my have attributes including a START_DATE, and an
>END_DATE spanning over random amounts of TIME.
>
>I would like to create a partitioned based table and generate a
>SQL*QUERY to query this data based on a "START_TIME_QUERY"
>and "END_TIME_QUERY. (i.e. select all segments where start_time >= T1
>and end_time <= T2, where a record (or segment) spans over time)
>
>Now, the question is, "How do I partition this table??" and what time
>based sql can I use to select these records.
>
>tia
>
>--
>Neil Greene
>Earthlink Network, Inc.
>Senior Oracle Database Administrator
>email: neilgreene_at_earthlink.net
>
>
>Sent via Deja.com http://www.deja.com/
>Share what you know. Learn what you don't.
Received on Fri Sep 24 1999 - 05:05:59 CDT