Thanks a lot , Srini !
I have this feeling too.
Unfortunately , bouncing the instance is not an option at the moment (it's
up since March , 2001).
The other problem is , when i'll restart the instance , the statistics will
disappear , so i never know my hit ratio ;-(
-Original Message-
Hi Christopher !
SQL select to_char(100 * misses / (hits + misses), '9990.00') || '%'
miss_rate
2from ( select total_waits misses
3from sys.v_$system_event
4 where event = 'db file sequential read' ),
5 ( select sum(dbbget + conget - pread)
Hello Jonathan !
Thanks for reply !
First , i love your site and may i recommend it to all the listers :
http://www.jlcomp.demon.co.uk/#Index
Second, yes , U are right :
SQL col HITS for 999,999,999,999,999,999
SQL select sum(dbbget + conget - pread) hits from x$kcbwds
2where inst_id =
recipients of list ORACLE-L
Asunto: RE: negative value for buffer cache hit ratio
Hi,
Generally speaking.. HIT RATIOS does not give the true
picture always. I have seen databases performing
extremely good with 50% and bad databases with 99% hit
ratio.
You should never decide the database performance
Interesting. In 8.1.7 32-bit oracle binary, select * from v$type_size where
type = 'UB4' tells me it's 4 bytes. So it's 32 bit and the maximum is 4
billion.
Glad to see Jonathan on this forum.
Yong Huang
[EMAIL PROTECTED]
you wrote:
From: K Gopalakrishnan
Welcome Jonathan,
I think most
, but there are no wait events then your
database does not need tuning.
Rachel
From: Ramon Estevez [EMAIL PROTECTED]
Reply-To: [EMAIL PROTECTED]
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Subject: RE: negative value for buffer cache hit ratio
Date: Wed, 08 Aug 2001 06:27:11 -0800
.
If the cache hit ratio is low, but there are no wait events then your
database does not need tuning.
Rachel
From: Ramon Estevez [EMAIL PROTECTED]
Reply-To: [EMAIL PROTECTED]
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Subject: RE: negative value for buffer cache hit ratio
Date: Wed, 08 Aug
hit ratio is low, but there are no wait events then your
database does not need tuning.
Rachel
From: Ramon Estevez [EMAIL PROTECTED]
Reply-To: [EMAIL PROTECTED]
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Subject: RE: negative value for buffer cache hit ratio
Date: Wed, 08 Aug
Try this more accurate query:
select to_char(100 * misses / (hits + misses), '9990.00') || '%' miss_rate
from ( select total_waits misses
from sys.v_$system_event
where event = 'db file sequential read' ),
( select sum(dbbget + conget - pread) hits
Andrey,
I am not into cache hit ratios, but got curious about your negative value.
So I casually searched Metalink for 'negative value for hit ratio'. And got
a hit on 'MROC: I am receiving a negative Buffer Cache hit ratio' thread.
You may want to check it out, there seems to be a new
1.- Go to this link of Steve Adam's site:
http://www.ixora.com.au/scripts/cache.htm
You gonna find a script to estimate the miss rate of
the buffer cache based on the assumptions that certain
operations are not well compute by statistics as
direct path operations.
2.- From 8.X you should rely
Ran this query under 8i(HP-UX 11)
SQL show user
USER is SYS
SQL select to_char(100 * misses / (hits + misses), '9990.00') || '%'
miss_rate
2from ( select total_waits misses
3from sys.v_$system_event
4 where event = 'db file sequential read' ),
5
Kirti, Audrey,
I was under the impression that negative numbers are caused by the numbers
wrapping around i.e., the database has been up for a while and the
statistics pile up and eventually exceed the defined format (e.g. value of
1000+ for a format of 9(3) - to use old COBOL representation !).
Nothing is wrong in your database.
You don't have the required X_$ views created. Change
X_$ to X$ or create X_$ views as select * from X$
views.
=
Have a nice day !!
Best Regards,
K Gopalakrishnan,
Bangalore, INDIA.
It is possible that after 4 months your stats
have wrapped around the ( ? 64 bit ?) limit
value for your platform. Check the actual
values from v$sysstat to see if some of them
have gone negative or appear to be
'counting backwards'.
Jonathan Lewis
Seminars on getting the best out of Oracle
I assume you have 8.1.7, since 8.1.7 they changed things back to how it was
done in 7.3.4.
This script works for 7.3.4 and 8.1.7/9.0.1
select to_char(100 * misses / (logical - physical + misses), '9990.00') ||
'%' miss_rate
from ( select total_waits misses
from sys.v_$system_event
Welcome Jonathan,
I think most of the counter are limited by ub4maxval
and that makes the negative hit ratio.
Welcome again !
--- Jonathan Lewis [EMAIL PROTECTED]
wrote:
It is possible that after 4 months your stats
have wrapped around the ( ? 64 bit ?) limit
value for your
Christopher
Thanks. This script ran from 7.3.4 to 8.1.6 with following result...
I hope misses less than 1% is not bad...Any comment
SQL select to_char(100 * misses / (logical - physical + misses), '9990.00')
||
2 '%' miss_rate
3from ( select total_waits misses
4from
Hi,
My misses is
MISS_RATE
-
0.68%
Is that good or bad, Oracle 8.0.5 Standard Edition, Windows 2000.
Is there a 8.1.7 Standard Edition Version ?
Ramon Estevez
[EMAIL PROTECTED]
---
Christopher
Thanks. This script ran from 7.3.4 to 8.1.6 with following result...
I hope misses
Hi,
Generally speaking.. HIT RATIOS does not give the true
picture always. I have seen databases performing
extremely good with 50% and bad databases with 99% hit
ratio.
You should never decide the database performance based
on hit ratio and 90% hit ratio does not mean that 90%
of the data is
20 matches
Mail list logo