From: "Harald Henkel" Newsgroups: comp.databases.oracle.server Subject: Re: NT PERFMON Answer and question Date: Fri, 2 Feb 2001 18:48:53 +0100 Lines: 70 Message-ID: <95es0n$gsb95$1@ID-56141.news.dfncis.de> References: <95chdb$561@NWNEWS.PCT.EDU> <95cnmp$gph75$1@ID-56141.news.dfncis.de> <95cqbt$h6g10$1@ID-56141.news.dfncis.de> <3a7ac8c4@news.iprimus.com.au> NNTP-Posting-Host: pec-83-112.tnt5.m2.uunet.de (149.225.83.112) X-Trace: fu-berlin.de 981136216 17706277 149.225.83.112 (16 [56141]) X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Hello Howard. "Howard J. Rogers" schrieb im Newsbeitrag news:3a7ac8c4@news.iprimus.com.au... > "Harald Henkel" wrote in message > news:95cqbt$h6g10$1@ID-56141.news.dfncis.de... > > Of course - a heavy loaded DB (or any server) will access its disks all > > the time. > > It's just some "strange" feeling that this frequent and IMHO unnessessary > > access - shouldn't be normal. > > Is this for security / recovery ? > In a word, yes. > The idea is that no-one in their right mind would want to go three seconds > without a minor checkpoint, on the grounds that -since you're using Oracle- > this is likely to mean a million and half transactions having to be repeated > if you ever have an Instance Failure. (If you were using SQL Server, > clearly there would only be three-and-a-half transactions, so who would > care?) > > Now: the good news is that in the training documentation, it clearly states, > time and time again (and has done so since version 8.0) that this 3 seconds > is the "default" time for minor checkpoints.... and we all know that > defaults can be changed, right? > > The bad news is that I haven't a clue off the top of my head how to change > the said default. Other than to make sure that log_checkpoint_timeout is > set to 0 and log_checkpoint_interval is set to something utterly huge. > > But then it occurs to me that worrying about a wee bit of disk activity is a > bit like the dinosaurs in 65 million BC worrying about how warm it was > getting... Thanks for the additional insight. When I know, what's it used for, I don't have a problem with it. I just wonder what Oracle is writing every 3 seconds on a system, where nothing happens. If it was in a stable state 3 seconds ago, it will be now, too, if nothing had happend inbetween. How is this related to archive logging ? If archive logging would be active (at the moment it isn't on that installation) I thought the DB would be in a stable state, when archive files are switched. And on a heavy duty DB this might well be about every 3 seconds. Well, of course in a pause - lets say after working stopped at might, this might not be true and the archive file not closed. If then the database might get corrupted, the transactions in the not closed archive file might be lost. If it closes the file every 3 seconds, that could be prevented. Is that, what it's for ? Harald Henkel