RE:db_file_scattered_read high waits

From: <Paula_Stankus_at_doh.state.fl.us>
Date: Tue, 16 Nov 2004 05:54:42 -0500
Message-ID: <E518A02B320C8D4E97D99A8C546480210251C05E@dohsmailhq01.doh.ad.state.fl.us>

Okay,

There have been a lot of deletes/inserts in an environment resulting in a high db_file_scattered_read wait I believe on a small table when doing a full table scan and retrieving a specific row. Basically to retrieve one row from this table I had to scan 377 blocks according to the trace each of 16K. =20

So I have some questions:

-I am using LMT with uniform extents and have a few 100 extents - my
understanding is that the extent sizing shouldn't play a part in this issue.

-I believe the large amount of deletes/inserts have. How to minimize
the deletes/inserts performance impact? At this point it appears that I would have to rebuild the table. However, in the future there will be additional deletes/inserts and I shouldn't have to keep rebuilding the table to ensure good performance. =20

-I am using automatic segment management.
--

http://www.freelists.org/webpage/oracle-l Received on Tue Nov 16 2004 - 04:50:08 CST