Re: performance issue on select count(*)

2003-10-28 Thread Linda Wang
: performance issue on select count(*) Date: Mon, 27 Oct 2003 10:34:59 -0800 Linda, I am guessing that since your table is partitioned on an unspecified date column, that the index on TID is either LOCAL or non-partitioned (i.e. GLOBAL). If it is LOCAL (you would have had to specify the keyword

Re: performance issue on select count(*)

2003-10-28 Thread Binley Lim
The symptom suggests caching is a big factor here - most likely block-buffers. Contrary to ?current? popular beliefs, BCHR is still a very relevant performance indicator - either being very high, or being too low - both of which gives a good indication of something that needs to be looked at.

Re: performance issue on select count(*)

2003-10-28 Thread Mladen Gogala
So, what exactly is indicated by a high or low hit rate? What, exactly, is high and what do you consider low? What HR are you talking about? This would be the infamous BCHR: select 'bc_hit_ratio' ratio,( sum(decode(name, 'consistent gets',value,0)) + sum(decode(name,'db block gets', value,0))

Re: Re: performance issue on select count(*)

2003-10-28 Thread ryan_oracle
: Re: performance issue on select count(*) So, what exactly is indicated by a high or low hit rate? What, exactly, is high and what do you consider low? What HR are you talking about? This would be the infamous BCHR: select 'bc_hit_ratio' ratio,( sum(decode(name, 'consistent gets',value,0

Re: performance issue on select count(*)

2003-10-28 Thread Jared . Still
To:Multiple recipients of list ORACLE-L [EMAIL PROTECTED] cc: Subject:Re: performance issue on select count(*) So, what exactly is indicated by a high or low hit rate? What, exactly, is high and what do you consider low? What HR are you talking about? This would

Re: Re: performance issue on select count(*)

2003-10-28 Thread Binley Lim
have to interpret it. that is about it. Anyone who uses it for anymore than that is wrong. From: Mladen Gogala [EMAIL PROTECTED] Date: 2003/10/28 Tue PM 12:09:34 EST To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED] Subject: Re: performance issue on select count(*) So, what

RE: performance issue on select count(*)

2003-10-27 Thread Stephane Faroult
Linda, I guess that the key word is 'partition'. This type of query should not require to access the table if (hopefully) tid is indexed. If the index on tid is also partitioned, all index partitions have to be searched. My feeling is that in such a case what should run faster is some

RE: performance issue on select count(*)

2003-10-27 Thread Linda Wang
PROTECTED] Reply-To: [EMAIL PROTECTED] To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED] Subject: RE: performance issue on select count(*) Date: Mon, 27 Oct 2003 05:49:24 -0800 Linda, I guess that the key word is 'partition'. This type of query should not require to access the table

Re: performance issue on select count(*)

2003-10-27 Thread Tim Gorman
Linda, I am guessing that since your table is partitioned on an unspecified date column, that the index on TID is either LOCAL or non-partitioned (i.e. GLOBAL). If it is LOCAL (you would have had to specify the keyword, as it is not the default), then you will be performing indexed RANGE scans