The tablespace parameters are
EXTENT MANAGEMENT LOCAL UNIFORM SIZE 256K
SEGMENT SPACE MANAGEMENT AUTO
Db block size is 16k
And db_file_nultiblock_read_count is unset in the spfile and is 64 in
From: Matt Schouviller
Sent: Wednesday, August 18, 2010 2:47 AM
To: kennethnaim_at_gmail.com; email@example.com
Subject: RE: Single block reads instead of multiblock reads
Can you provide the configuration for the tablespace(s) that this table is
in? Also please provide the block size and the value of the
db_file_multiblock_read_count init parameter.
On Behalf Of Kenneth Naim
Sent: Tuesday, August 17, 2010 10:17 PM
Subject: Single block reads instead of multiblock reads
I have a query that needs to read 25% of a 100gb (116 million row, 600
columns) table which has been recently analyzed. The database is 18.104.22.168,
sga is 35gb, pga 15gb, solaris, 16 cpus, 16k blocksize.
The query only accesses one table. When run single threaded or in parallel
the query runs for hours with the top wait event being db file sequential
read not sequential reads with the p3 showing 1 block. I have checked the
explain plan, rowsource in a tkrpof-ed trace file, the v$sql_plan all show
full table scan. I've specified a full hint with the correct alias and I'm
at a loss on why I'm not seeing a multiblock read.
I've even simplified the case to be ctas * from big_table where
rownum<1000000 and it still does single block reads.
Anyone has ideas to why this could be happening?