Re: RAC partitioning question
From: Greg Rahn <greg_at_structureddata.org>
Date: Tue, 31 Jan 2012 09:24:31 -0800
Message-ID: <CAGXkmisMm74tkd8tfDrxwjH+OQm+PYjPWjs3_gWgjb+qJgd3sQ_at_mail.gmail.com>
Any reason not to just add the date/timestamp column into the PK and make it two columns? This would result in allowing the index to be local (the date col is the partition key col) without any modification to the current table definition.
On Tue, Jan 31, 2012 at 8:42 AM, Tim Gorman <tim_at_evdbt.com> wrote:
Date: Tue, 31 Jan 2012 09:24:31 -0800
Message-ID: <CAGXkmisMm74tkd8tfDrxwjH+OQm+PYjPWjs3_gWgjb+qJgd3sQ_at_mail.gmail.com>
Any reason not to just add the date/timestamp column into the PK and make it two columns? This would result in allowing the index to be local (the date col is the partition key col) without any modification to the current table definition.
On Tue, Jan 31, 2012 at 8:42 AM, Tim Gorman <tim_at_evdbt.com> wrote:
> Jed,
> Why not get rid of the sequence-generated PK column and instead make
> another NUMBER column the PK, itself generated both from the needed
> timestamp appended to a sequence generated data value to ensure it's
> uniqueness? If you have that, then you can RANGE partition on that
> NUMBER value according to your data manipulation requirements and also
> have a LOCAL partitioned index so that you have no GLOBAL index issues.
>
>
-- Regards, Greg Rahn | blog <http://bit.ly/u9N0i8> | twitter <http://bit.ly/v733dJ> | linkedin <http://linkd.in/gregrahn> -- http://www.freelists.org/webpage/oracle-lReceived on Tue Jan 31 2012 - 11:24:31 CST