Thank You Jonathan for the trick of comparing usage bucket count with the
optimizer stats collection history.
On Fri, Nov 27, 2020 at 3:36 PM Jonathan Lewis <jlewisoracle@xxxxxxxxx>
wrote:
You can almost guarantee that there won't be any new option in 19.3 when
the latest strategy only appeared in 12.2. It's disappointing that the
stats collection usage hasn't been disabled.
However, one of the usage buckets is "> 1000", so for any big indexes that
are not needed you could take the view that seeing that increment once per
day, or once per week is an indication that it's a stats collection usage,
and you could correlate this with the the optimizer stats history of
operations to check that each usage correlates with stats collection.
Regards
Jonathan Lewis
On Fri, 27 Nov 2020 at 07:50, Lok P <loknath.73@xxxxxxxxx> wrote:
yes, but i was thinking as we are in version 19C, and i see in the blog
stating, the index usage is now by default taken care by Oracle starting
version 12.2 and the details gets flushed into DBA_INDEX_USAGE view. And no
need to set the index monitoring ON specifically as it used to happen in
11.2.
But the issue is, when I tried collecting stats on the index , even then
the DBA_INDEX_USAGE was getting populated which is wrong. It should only be
populated when it really gets accessed for fetching data like SELECT
queries etc. So wanted to understand , if any other way through which we
will get the index usage correctly(may be new in 19.3)? Also foreign key
access is not captured in this case, and dropping those indexes will be
dangerous.
https://oracle-base.com/articles/12c/index-usage-tracking-12cr2
On Fri, Nov 27, 2020 at 3:59 AM Mladen Gogala <gogala.mladen@xxxxxxxxx>
wrote:
Hi Lok,
You can turn on monitoring for index, something like "ALTER INDEX PK_EMP
MONITORING USAGE". The usage will be in V$OBJECT_USAGE. Unfortunately,
there is no owner in that table, so you need to log in as the index owner.
Regards
On 11/26/20 3:19 PM, Lok P wrote:
We have just moved from Oracle version 11.2.0.4 to 19.3. And also we
were on HP and now we moved to Exdata-X5. And the team is asking to verify
if some of the indexes are actually not in use as we are in Exadata and
thus we can drop them afterwards. So was trying to understand the reliable
way to achieve the list of unused indexes?
As per my understanding 11.2.0.4 was having option for setting the
monitoring On for the specific indexes, to see its usage from
dba_object_usage, but that was just a "YES" or "NO" flag which was not of
much help(as amount of usage is also not captured there) and also stats
gathering on the index was making the usage flag to YES which is wrong,
also indexes on foreign key were not getting picked up as USED in cases.
But i see starting from version 12.2+ , Oracle is by-default capturing
all the information in a new view DBA_INDEX_USAGE, so wanted to understand
from experts if this is safe and reliable way to collect index usage
information or if we have anything new on 19C more reliable for finding
index usage?
--
Mladen Gogala
Database Consultant
Tel: (347) 321-1217
