Accessing Tables using global index rowid is taking too long.

Question

I am using Oracle 10.2.0.4.0. I am having tables which are ~100+GB in size which is range partitioned on a date column and hash subpartitioned on primary key column. And accessing these tables using global index row id using idex range scan( size of those indexes are ~7+ GB) taking ~5+ mins which results in ~1000 rows as out put. So is this means there is some structural problem with tables/indexes or its something issue at block level? How to dig more into this issue?

Answer 1

Hi,
930254 wrote:
I am using Oracle 10.2.0.4.0. I am having tables which are ~100+GB in size which is range partitioned on a date column and hash subpartitioned on primary key column. And accessing these tables using global index row id using idex range scan( size of those indexes are ~7+ GB) taking ~5+ mins which results in ~1000 rows as out put. So is this means there is some structural problem with tables/indexes or its something issue at block level? How to dig more into this issue?once you get a chance to run the query and collect some diagnostic info, here are the questions that you should ask yourself:
1) where is the 5 mins spent? doing index range scan itself, doing table index access by global rowid, waiting for something, doing something else?
2) how many reads are there and what kind (logical, physical, direct path)? per the entire query and per index range scan
3) is there any aggregation in the query? if there is, then 1000 rows output doesn't mean anything -- you could have to access millions or billions of rows to produce a 1000 row output in an aggregate query
The best tool to answer these questions is dbms_xplan.
Best regards,
Nikolay