Re: When does LGWR flush the ebuffer?

From: Richard Foote <richard.foote_at_bigpond.com>
Date: Wed, 23 Apr 2003 23:41:48 +1000
Message-ID: <Wwwpa.20979$1s1.316824@newsfeeds.bigpond.com>

"Ryan" <rgaffuri_at_cox.net> wrote in message news:Lrjpa.149613$yh1.9548056_at_news1.east.cox.net...

> someone on a listserv I get posted that when you issue a commit DBWR
> activates and writes to disk. Is this accurate? I thought DBWR was only
> flagged and that a checkpoint determined when it was written to the
datafile
> because its more costly to write to a datafile then to a redo log?
>

Hi Ryan,

No, it's not accurate, the commit wakes up LGWR, not DBWRs ...

Nor is it accurate to say that a checkpoint (necessarily) causes the block to be written to disk. There are a number of "events" that "wakes" up the DBWRs such as 3 seconds, dirty list thresholds, etc.

Cheers

Richard Received on Wed Apr 23 2003 - 08:41:48 CDT