Xref: alice comp.databases.oracle.server:47899 Path: alice!news-feed.fnsi.net!newspump.monmouth.com!newspeer.monmouth.com!dispose.news.demon.net!demon!bullseye.news.demon.net!demon!news.demon.nl!demon!sybrandb.demon.nl!not-for-mail From: "Sybrand Bakker" Newsgroups: comp.databases.oracle.server Subject: Re: Hash_join_enabled: what if equals FALSE Date: Fri, 30 Apr 1999 23:02:47 +0200 Organization: Bentley Systems Europe Message-ID: <925506032.13893.0.muttley.d4ee154e@news.demon.nl> References: X-Trace: news.demon.nl 925506032 muttley:13893 NO-IDENT sybrandb.demon.nl:212.238.21.78 X-Complaints-To: abuse@demon.net X-Newsreader: Microsoft Outlook Express 4.72.3110.5 X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 Lines: 39 Hi Michel Joins can be performed in one of three manners nested loops (retrieve one row from a master table, subsequently retrieve all rows from the corresponding detail table) merge-join (join is performed by executing a sort, all applicable rows are buffered in memory) and hash-join (a hash key is being calculated and this key is used for sorting purposes. In merge-join the complete select expression is sorted) Turning the parameter off simply means you can't use it. Now this hash-join mechanism might (?!) be supported by the cost based optimizer only. I have a feeling if you are running rule based (this happens with optimizer_mode = choose and if you don't have statistics in the dictionary) it won't be used anyway. Hth, Sybrand Bakker, Oracle DBA Michel Gauthier wrote in message ... >Hi all, > >After having a few ORA-600 [15851] [6] [6] etc... I searched the Oracle Web >support to find out that they sometimes suggest to turn off the >HASH_JOIN_ENABLED parameter. This parameter seems to appear with 7.3 and I'm >not familiar with it. Anyway, I turned it off, and it seems to actually >work. Fine. > >Unfortunately, that does not say what this parameter is useful to, neither >which are the impacts to turn it off (except it workaround my problem!). I >surfed the Oracle doc and DEJANEWS without success. > >Can anybody tell us about this parameter. That would be nice. > >TIA. > >Michel Gauthier > >