Re: Preventing delayed block cleanout due to loading a datawarehouse
Date: Wed, 20 Apr 2011 02:37:56 -0700 (PDT)
Message-ID: <cea75c55-95ca-4fbf-b631-eb5ec5ad4acd_at_dn9g2000vbb.googlegroups.com>
On Apr 19, 2:51 pm, Donatello Settembrino <donatello.settembr..._at_gmail.com> wrote:
> Exactly, that's right
By the way - we digress - my comment about PCTFREE is irrelevant in this case since Oracle will stuff this row into a single block no matter what the PCTFREE setting is as long as the row fits into the block.
And: You don't have chained rows but your test case is flawed: A select count(*) never reports "table fetch continued row" because it does not have to visit the columns but only accesses the row entries in the row directory. So even with chained rows your test will not report any "table fetch continued row" statistics. You would need to count/access a column that is part of the "chained" row pieces - and it would have to be nullable, otherwise recent versions of Oracle will transform a count(col) to a count(*).
Randolf Received on Wed Apr 20 2011 - 04:37:56 CDT