Path: news.easynews.com!easynews!peer1-sjc1.usenetserver.com!usenetserver.com!cyclone.bc.net!newsfeed.direct.ca!look.ca!wn1feed!worldnet.att.net!204.127.198.203!attbi_feed3!attbi.com!rwcrnsc54.POSTED!not-for-mail From: "Jim Kennedy" Newsgroups: comp.databases.oracle.server References: <5KZX7.36613$Q66.129567@NewsReader> Subject: Re: Oracle and SQL Server Lines: 28 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4807.1700 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700 Message-ID: NNTP-Posting-Host: 12.224.242.123 X-Complaints-To: abuse@attbi.com X-Trace: rwcrnsc54 1009910955 12.224.242.123 (Tue, 01 Jan 2002 18:49:15 GMT) NNTP-Posting-Date: Tue, 01 Jan 2002 18:49:15 GMT Organization: AT&T Broadband Date: Tue, 01 Jan 2002 18:49:15 GMT Xref: easynews comp.databases.oracle.server:129333 X-Received-Date: Tue, 01 Jan 2002 11:53:50 MST (news.easynews.com) Good points. Also I believe the default in the sql query tool is autocommit. Jim "Kendall" wrote in message news:pan.2001.12.31.10.31.05.466.8939@yahoooo.com... > In article <5KZX7.36613$Q66.129567@NewsReader>, "Garrick Bigwood" > wrote: > > > Hello, > > > Looking at the trace files one of the problems with > > Oracle seems to be the amount of physical writes done to datafiles and > > rollback files(also redo of course), when I look at the SQL Server trace > > files there does'nt to be any writing done. > > Just to clarify, are you doing a COMMIT after each update? I don't know > exactly how SQL server handles these, but you may be right about delayed > i/o. A commit should generate some kind of write to disk. > > Oracle only requires a write to the redo log on commit; any writes to > datafiles are from cache flushing, or a checkpoint. If your SGA is too > small you're obviously going to see more I/O. Check the SGA and SQL > Server's cache size - they should be the same size for a fair test. > > Also, how many indexes are on the table/column? Is the column NULL before > update?