Xref: alice comp.databases.oracle.server:45529 Path: alice!news-feed.fnsi.net!news.maxwell.syr.edu!dispose.news.demon.net!demon!news-lond.gip.net!news-raspail.gip.net!news.gsl.net!gip.net!grolier!club-internet!not-for-mail From: Stephane Faroult Newsgroups: comp.databases.oracle.server Subject: Re: ------PLEASE HELP------------- Date: Fri, 09 Apr 1999 19:43:25 -0700 Organization: Oriole Corporation Message-ID: <370EBACD.4148@oriolecorp.com> References: <370DE79C.28DCC6E2@india.hp.com> Reply-To: sfaroult@oriolecorp.com Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Trace: front6.grolier.fr 923687214 29741 194.158.113.251 (9 Apr 1999 19:46:54 GMT) NNTP-Posting-Date: 9 Apr 1999 19:46:54 GMT X-Mailer: Mozilla 3.04 (Win16; I) To: "G.H VITTAL" Lines: 38 > 1) Redolog files - should it increase ? Only if you notice 'Checkpoint not complete' messages in your alert.log file. > 2) Redolog buffers - Should it increase ? Once again, it depends on what you get in statistics (V$SYSSTAT). Compare 'redo log space requests' to 'redo entries'. Increase your redolog buffers if its more than a few percents. Another indicator is the ratio of 'redo writes' to 'redo sync writes' - above 2 I would increase the redo log buffers. > 3) Rollback segments - The records are always commited for every 100 > records. Whether should i increase rollback segments ? You need probably not increase them in size, individually, but you should increase their number by at least a factor 2. Look in V$WAITSTAT. If you have a noticeable number of waits for 'undo header blocks' it means you do not have enough rollback segments. > Obviously I have to increase the size of the data file for the > tablespace because of the number of records > increased from 2 million to 6 million. Will the size of the data file > for the > database should be increased or automatically taken care ? If you are running Oracle 7.2 or above, datafiles can be put in 'autoextend' mode - they will automatically increase in size when required. I may be a little old-hat, but I am rather in favour of not using this and allocating datafiles manually. I think that replacing an Oracle 'Cannot allocate extent ...' message by a Unix 'file system full' message is fudging rather than solving the problem. Better to estimate volumes beforehand and monitor it 'inside' Oracle. Just ask your systems engineer! -- Regards, Stéphane Faroult Oriole Corporation ------------------------------------------------------------------ http://www.oriolecorp.com, designed by Oracle DBAs for Oracle DBAs ------------------------------------------------------------------