AWR Sample Report
in [Oracle]
|
From: Steve Howard on 18 Nov 2008 16:29 On Nov 18, 1:38 am, raja <dextersu...(a)gmail.com> wrote: > Hi, > > I have another AWR Report generated with 1hr interval. ??? > > AWR Report : > Snap Id Snap Time Sessions Cursors/Session > Begin Snap: 10739 13-Nov-08 09:00:45 105 58.2 > End Snap: 10763 13-Nov-08 15:00:12 91 75.2 ???
From: Steve Howard on 18 Nov 2008 16:36 On Nov 18, 1:38 am, raja <dextersu...(a)gmail.com> wrote: What is the issue with all of this, any way? It looks like you had an average of 30% USER CPU for the duration of the snap (92,000 seconds of USER CPU, with 14 CPU's over six hours of available CPU to be used). That's not too bad? Is an SLA not being met? Also, why did you set these ? init.ora Parameters Parameter Name value >> _newsort_enabled FALSE >> _optim_peek_user_binds FALSE
From: joel garry on 18 Nov 2008 18:48 On Nov 18, 1:36 pm, Steve Howard <stevedhow...(a)gmail.com> wrote: > Also, why did you set these ? > > init.ora Parameters > Parameter Name value > > > > >> _newsort_enabled FALSE > >> _optim_peek_user_binds FALSE Can't speak for the OP, but at least one vendor of mine is magic- bulleting _optim_peek_user_binds jg -- @home.com is bogus. http://www.rottenneighbors.com
From: Steve Howard on 18 Nov 2008 21:52 On Nov 18, 6:48 pm, joel garry <joel-ga...(a)home.com> wrote: > On Nov 18, 1:36 pm, Steve Howard <stevedhow...(a)gmail.com> wrote: > > > Also, why did you set these ? > > > init.ora Parameters > > Parameter Name value > > > >> _newsort_enabled FALSE > > >> _optim_peek_user_binds FALSE > > Can't speak for the OP, but at least one vendor of mine is magic- > bulleting _optim_peek_user_binds > > jg > -- > @home.com is bogus.http://www.rottenneighbors.com We actually used it once for Oracle Text queries that would misbehave. We *always* wanted to use the Text index, but sometimes the optimizer would build a different cursor based on the bind that was peeked. We literally had over 10,000 child cursors for a given statement. The soft parsing killed us. We turned off bind peeking, and the problem went away. Six months later, it bit us in the rear for something else, though. Regardless, that would still be the first question I would ask, though.
From: raja on 19 Nov 2008 01:14
Hi, Thanks for ur comments. I trying to learn, how to interpret AWR Report. Previously i had an awr report here ( pls see the previous posts of this thread ), which charles, steve and sybrand helped to give me some tips so that i can investigate more on it. Similarly i have another AWR Report here with 1 hr interval ( having this, as a place to make a study/analysis of awr report ) Just want to study this report too. I dont know where to start from / where to look into first ( hot spots in awr report to look into ), thats y pasted the awr report here for discussion. Still cant understand what steve Howard and joel garry, try to comment on ? Please explain. Steve Howard, May i know how you calculated the below ones ... "It looks like you had an average of 30% USER CPU for the duration of the snap (92,000 seconds of USER CPU, with 14 CPU's over six hours of available CPU to be used)." Joel Garry, Can you explain again on what you said ? Thanks in Advance. With Regards, Raja. |