Path: dp-news.maxwell.syr.edu!spool.maxwell.syr.edu!drn.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.media.kyoto-u.ac.jp!news1.optus.net.au!optus!newsfeeder.syd.optusnet.com.au!news.optusnet.com.au!not-for-mail Date: Mon, 13 Dec 2004 14:08:27 +1100 From: "Howard J. Rogers" User-Agent: Mozilla Thunderbird 0.9 (X11/20041127) X-Accept-Language: en-us, en MIME-Version: 1.0 Newsgroups: comp.databases.oracle.server Subject: Re: DBWR Tuning References: <1102906957.981203.92230@c13g2000cwb.googlegroups.com> In-Reply-To: <1102906957.981203.92230@c13g2000cwb.googlegroups.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Lines: 38 Message-ID: <41bd07a6$0$6539$afc38c87@news.optusnet.com.au> NNTP-Posting-Host: 203.164.5.49 X-Trace: 1102907303 6539 203.164.5.49 Xref: dp-news.maxwell.syr.edu comp.databases.oracle.server:231438 wagen123@yahoo.com wrote: > Oracle9i 9.2.0.5 (64 bit) > Solaris 2.9 (64 bit) > RAID 0+1 > OLTP db > > Ref: Steve Adams - The Mysteries of DBWR Tuning > > We are trying to address the question "should we have multiple DBWR > processes?" > > My understanding is that - If one does not see any wait events related > to "free buffer waits" OR "write complete waits", DBWR tuning is not > required and multiple DBWR processes is not required. > Open to suggestions. > > Thanks > wagen > Checkpoint incomplete messages in the alert log can indicate a sluggish DBWR (it can also be a sluggish ARCH, as well... ) Same for redo retries and redo allocation waits. Could be silly sized redo logs, could be too few log groups, could be poor ARCH performance. Could be sluggish DBWR. Usually, you'd check other wait events to be certain which of the possibles actually applies. So, for example, if I have redo allocation problems, AND free buffer waits, I know faster hard disks or more DBWRs could help. If I have checkpoint incomplete messages, but no free buffer waits, then I'd suspect ARCH was mucking around. On the other hand, free buffer waits might not mean more DBWRs. It might mean 'lose the RAID5'. Just as a for-example! Regards HJR