Are hash sub-partitions balanced?
Date: Thu, 2 May 2013 10:51:50 -0400
Message-ID: <CAE-dsOLh+yCXXnZ0mHq6iACyLD1zOw9HbHPa8xgP-XuUwukm5Q_at_mail.gmail.com>
Database 11.2
If I have highly skewed data, will oracle balance out the sub-partitions? I won't have real production data to measure the skew until I actually deliver this. So we are trying to simulate data and do inserts. I don't know how accurate our estimates of the skew are.
I am wondering if I may get some very large sub-partitions and some very small ones. There has been some talk that we should use list sub-partitions, but this would be a huge hassle. There are 700-800 possible values and these can change. So I would have to have maintenance to handle 'splits'. My maintenance windows are very limited.
That being said, this will be a high activity database with 600+ inserts/second 24x7 and a similiar number of queries. There will be a large number of updates, but none on the partitioned tables. My SLA on this is measured in fractions of a second.
just curious if anyone tested this? I
also does anyone know if its possible to use templates to give hash-subpartions naming conventions. This is just to make it easier for me to analyze the data and do rollups. I was thinking that subpartiions names that repeat such as
1A
1B
so for the next date partion the sub-partiion would be
2A
2B
It looks like oracle uses a sequence for naming sub-partitions, so I should be able to rollup using a modulus, but this is a bit of a hassle.
-- http://www.freelists.org/webpage/oracle-lReceived on Thu May 02 2013 - 16:51:50 CEST