Re: serial direct reads
From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Thu, 4 Sep 2014 11:04:47 +0100
Message-ID: <yf6dnXDn69NypZXJnZ2dnUVZ8kadnZ2d_at_bt.com>
Date: Thu, 4 Sep 2014 11:04:47 +0100
Message-ID: <yf6dnXDn69NypZXJnZ2dnUVZ8kadnZ2d_at_bt.com>
The lower limit is 2% , not 10% - though that's probably only correct for 11.2.0.x
http://jonathanlewis.wordpress.com/2011/03/24/small-tables/ http://afatkulin.blogspot.co.uk/2012/07/serial-direct-path-reads-in-11gr2-and.html
-- Regards Jonathan Lewis http://jonathanlewis.wordpress.com/all-postings Author: Oracle Core (Apress 2011) http://www.apress.com/9781430239543 "vsevolod afanassiev" <vsevolod.afanassiev_at_gmail.com> wrote in message news:619c0a5d-3325-405c-971f-15e62fdb8638_at_googlegroups.com...Received on Thu Sep 04 2014 - 12:04:47 CEST
| Real-life example:
| 11.2.0.3 on AIX
| DB_CACHE_SIZE = 4GB, DB_BLOCK_SIZE = 8K, so DB_CACHE_SIZE = 524,288
blocks
| 10% of DB_CACHE_SIZE = 52,429 blocks
| _SMALL_TABLE_THRESHOLD = 10,257.
| Table has DBA_TABLES.BLOCKS = 12,232
| _DIRECT_READ_DECISION_STATISTICS_DRIVEN = TRUE
|
| As table size is much smaller than 10% of DB_CACHE_SIZE one would expect
that full table scan will be using buffered reads. However it is using direct reads (top wait event = 'direct path read'). Any idea why?