Re: buffer cache structure in SGA

From: Orlando L <>
Date: Tue, 3 Apr 2018 18:14:38 -0500
Message-ID: <>


The question is not how much to reduce it to, the paper gives an idea on the numbers. The question is how to reduce it. I feel that the answer to the question may be quite big in the form of a book that deals with several possible cases, possibly it involves a good understanding of the SQL statements, oracle internals, etc. But still any pointers will be useful.


On Tue, Apr 3, 2018 at 2:22 AM, Andy Sayer <> wrote:

> “Anyway, based on the paper, I started searching on ways to reduce
> logical IOs. Any pointers on that are welcome too. ”
> How much do you want to reduce it to? Why do you think your current levels
> are too high? Have you identified a problem?
> Once you have confirmed you don’t suffer from compulsive tuning disorder
> and you have a problem that is caused by too many LIOs. Look at where they
> are coming from, use your instrumentation to check which processes are
> responsible for the most LIO during times that it matters. Then tune the
> process so that it doesn’t require as much - this may be straight forward
> SQL tuning like adding appropriate indexes, helping the optimizer choose
> the correct path, filtering down earlier in joins. It could be rethinking
> about what your process should be doing, should it really be populating a
> GTT with all rows from table_x? Etc
> Hope that helps,
> Andrew

Received on Wed Apr 04 2018 - 01:14:38 CEST

Original text of this message