RE: db buffer cache advisory clarification

  • From: "Luca Canali" <Luca.Canali@xxxxxxx>
  • To: <st.anderson@xxxxxxxxx>
  • Date: Wed, 29 Mar 2006 12:11:20 +0200

Hi Stephen,
 
As for your question on consistency of the report for small cache
values, I think you still have a formatting problem. If you take row N.3
for example, the column "Est Physical Read Factor" tells you that the
value there is 1/0.02=50 times bigger than what is reported in row N.4,
that is the cache advisor expects 50 billion physical reads (your
sqlplus format  string handles numbers up to 10 billion)
Your next problem is to make any sense of those huge numbers..
Cheers,
Luca
 
 
________________________________

From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Stephen Anderson
Sent: Wednesday, March 29, 2006 11:28 AM
To: jungwolf
Cc: oracle-l@xxxxxxxxxxxxx
Subject: Re: db buffer cache advisory clarification


Hello,
 
Ran a 7 minute snap this morning - results are getting worse!  Physical
Reads are at 6278/s and Logical Reads are 70,844/s.
 
Buffer Pool Advisory for DB: MERLIN  Instance: MERLIN  End Snap: 31
-> Only rows with estimated physical reads >0 are displayed
-> ordered by Block Size, Buffers For Estimate 

        Size for  Size      Buffers for  Est Physical          Estimated
P   Estimate (M) Factr         Estimate   Read Factor     Physical Reads
--- ------------ ----- ---------------- ------------- ------------------

D             16    .3            1,985          0.07        124,118,645
D             32    .5            3,970          0.04         70,906,440
D             48    .8            5,955          0.02         29,259,452

D             64   1.0            7,940          1.00      1,783,067,416
D             80   1.3            9,925          0.99      1,757,988,089
...
D            256   4.0           31,760          0.85      1,523,232,446
D            272   4.3           33,745          0.85      1,507,672,223

Now we can see over 1 billion PIO's, so it cannot be a formatting thing.
The metalink note that Jared suggested merely chose to format it
differently. 

 
On 3/28/06, jungwolf <spatenau@xxxxxxxxx> wrote: 

        Stephen,
        
        This is for a 15 minute snapshot?  It seems a little strange to
me.
        >         Size for  Size      Buffers for  Est Physical
Estimated 
        > P   Estimate (M) Factr         Estimate   Read Factor
Physical Reads
        > --- ------------ ----- ---------------- -------------
------------------
        > D             64   1.0            7,940          1.00
937,436,311
        
        937436311/(15*60)=1041595.9 or over 1 million physical reads per
second.
        
        Maybe something else is going on?
        
        Steven

--
//www.freelists.org/webpage/oracle-l


Other related posts: