Kevin,
This would make sense.
Global indexes are not associated with individual partitions, therefore =
if
you work on one partition the whole index is invalidated (?). Better =
to use
local indexes, that way queries pertaining to the other partitions =
would not
be forced to do full table scans.
Thanks.
Patrice Boivin
Systems Analyst (Oracle Certified DBA)
Systems Admin & Operations | Admin. et Exploit. des syst=E8mes
Technology Services | Services technologiques
Informatics Branch | Direction de l'informatique=20
Maritimes Region, DFO | R=E9gion des Maritimes, MPO
E-Mail: boivinp_at_mar.dfo-mpo.gc.ca
> -----Original Message-----
> From: Toepke, Kevin M [SMTP:ktoepke_at_cms.cendant.com]
> Sent: Wednesday, November 29, 2000 12:36 PM
> To: Multiple recipients of list ORACLE-L
> Subject: RE: Local Prefix vs non-prefix indexes
>=20
> My take on that statement is they really meant:
> > "Local nonprefixed indexes provide more availability than prefixed
> indexes
> > when partition maintenance operations are carried out."
>=20
> "Local indexes provide more availability than global prefixed indexes =
when
> partition maintenance operations are carried out"
>=20
> Partition Maintenance Operations are adding, merging, splitting, =
dropping
> of
> individual partitions.
>=20
> If you use a global prefixed index, then when you perform a partition
> mainenance operation, the while index must be rebuilt to drop the =
dropped
> index blocks or update the "updated" rowids (in the case of a =
partition
> merge or split)
>=20
>=20
> > -----Original Message-----
> > From: Boivin, Patrice J [mailto:BoivinP_at_mar.dfo-mpo.gc.ca]
> > Sent: Wednesday, November 29, 2000 10:31 AM
> > To: Multiple recipients of list ORACLE-L
> > Subject: RE: Local Prefix vs non-prefix indexes
> >=20
> >=20
> > I sat through the Oracle8i New Features for Administrators=20
> > course, but I
> > don't know much.
> >=20
> > Here is my take on this:
> >=20
> > Prefixed indexes start with the partition key.
> > I don't know what Oracle means by "maintenance", but if it=20
> > relies on the
> > prefixed index to locate data blocks, that would tie up the=20
> > index during
> > maintenance.
> >=20
> > Non-prefixed indexes start with columns other than the=20
> > partition key column,
> > therefore Oracle wouldn't use those during maintenance on =
partitions.
> >=20
> > I am not sure that the other indexes pointing to that=20
> > partition would work
> > at all, however, it seems to me if Oracle shuffles data=20
> > blocks around in the
> > partition, then the indexes pointing to those blocks would be=20
> > invalidated.
> > Hence the rebuild index online feature... to speed up the=20
> > rebuilding of
> > indexes after work on partitions has completed.
> >=20
> > Now that I think about it, if it relies on the partition key=20
> > index to work
> > on that partition, the index would be invalidated as well,=20
> > wouldn't it?
> >=20
> > Bizarre.
> >=20
> > Where did you find that statement in the Oracle=20
> > documentation? Now I am
> > curious to know what they meant by "maintenance".
> >=20
> > Regards,
> > Patrice Boivin
> > Systems Analyst (Oracle Certified DBA)
> >=20
> > Systems Admin & Operations | Admin. et Exploit. des syst=E8mes
> > Technology Services | Services technologiques
> > Informatics Branch | Direction de l'informatique=20
> > Maritimes Region, DFO | R=E9gion des Maritimes, MPO
> >=20
> > E-Mail: boivinp_at_mar.dfo-mpo.gc.ca
> >=20
> <snip>
> --=20
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> --=20
> Author: Toepke, Kevin M
> INET: ktoepke_at_cms.cendant.com
>=20
> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> San Diego, California -- Public Internet access / Mailing =
Lists
> --------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
Received on Thu Nov 30 2000 - 06:11:08 CST