Difference between RTC Indices and Database indices
Hi,
We have repotool commands to create Text and workitem indices from RTC.
Now, in our Oracle Database, we have DB command to create indices.
My question is
1) Assume that we have an activity of splitting Bigfile table space to multiple tables . Following this, from the DB end, we will build indices in RTC
2) Again on RTC application end, do we have to run repotools command to rebuild text and workitem indices ? Can we not use the earlier built RTC workitem and text indices to look into the Database ?
How is running repotools command to build indices is different from building indices directly from Database ?
I'm looking at minimizing the time to build indices with RTC shutdown. If running indices in Database will be more than sufficient, then i would not prefer to run repotools command to build Text and Workitem indices.
2 answers
Please read https://jazz.net/library/article/1271 and the second part to understand what the indices do. Please read https://jazz.net/wiki/bin/view/Deployment/BackupCLM to understand how backup works.
You would only have to rebuild text indexes or database indexes in case of fatal issues - e.g. the loss or inconsistency of indexes. This would likely be due to some error indicating the indexes are compromised or due to support asking to do so.
As the application knows how the indexes are to be build, you should use the repotools command that supports it. You should not introduce indexes on DB level - unless support suggests to do so.
You would only have to rebuild text indexes or database indexes in case of fatal issues - e.g. the loss or inconsistency of indexes. This would likely be due to some error indicating the indexes are compromised or due to support asking to do so.
As the application knows how the indexes are to be build, you should use the repotools command that supports it. You should not introduce indexes on DB level - unless support suggests to do so.
Hi Ralph,
We did an exercise of splitting our oracle database created as Bigfile tablespace into multiple tables model. The DB tables were actually moved across different tablespace. As one of the completion steps, we had to build indices on the DB level. They were the default RTC table indices and nothing customized.
We were in a doubt whether the earlier RTC indices would be fine or have to rebuild them because of the changes in DB. We build the RTC indices again to avoid any performance problem that could arise on using the old RTC indices against the new DB structure and DB indices.
From the link that you have provided, my understanding is that the DB indices are independent of RTC indices and the decision on building DB indices should be purely based on DB changes. Also,building only RTC indices will not mean that DB indices need not be built.
My second question is based on the scenario that we have experienced.
Earlier, we used to get recommendations in DB monitoring suggesting us to build new indices to improve the performance of querying. However, we didn't build any new indices other than the default RTC indices that were part of the tables. What could be the problems on building custom indices in DB level ? Will there be any additional step required on RTC indices ?
