Are there any guidelines for the below ?
ixda-RA idx-RA da-RA RA-pgsused lchwaits
4398 29 27805 32188 55
Thanks,
Manoj Schroff,
Two Way TV Ltd, UK
+44(0)181 996 8221
ma...@mailhost.twowaytv.co.uk
http://www.twowaytv.co.uk
This shows that the Informix read-ahead logic has prefetched 4398 index
leaves, 29 index nodes, and 27805 data pages for a total of 32232
pages. You have actually used 32188 of those prefetched pages for a
Read-Ahead utilization rate of 99.86%. Pretty impressive. It shows
that you might benefit from increasing the Read-Ahead parameters in the
ONCONFIG file slightly. The latchwaits shows how often processes are
having to block (spinwaits are not counted here) because they cannot
get access to a shared memory resource. Unless the statistics were
cleared in the last minute or less (tbstat -z) 55 lchwaits is great.
I just cleared stats on one of our servers and about ten minutes later
we had 208 lchwaits with 432491 isamtot operations (just so you can
judge).
Art S. Kagel
Referring to page 40-86 in the ODS Admin Guide Vol2,
ixda-RA refers to the count of read-aheads from index leaves to data pages.
Put another way, its the number of read-ahead pages resulting from an
optimization path of INDEX PATH wherein the query reads the index and then
the data page.
idx-RA refers to the count if read-aheads traversing index leaves. Put
another way, it is the number of read-ahead pages resulting from an
optimization path of INDEX KEYwherein the entire query can be satisfied by
an index and does not require a page read to the table.
da-RA refers to the count of data-path-only scans. Otherwise known as
SEQUENTIAL SCAN.
In your example, 32,188 read-ahead pages were used out of the 32,232
read-ahead pages retrieved. That means only 44 pages were abandoned. That
is pretty good (99.86%). You may benefit from increases these parameters a
little. But watch for this percentage to drop off as it is an indication
that you are reading in too much and wasting I/O.
Now, in the event that I missed something, I'm sure my esteemed cohorts
here in c.d.i. will complete anything I missed.
Take Care,
Chuck.
+++-------------------------------------------------------------------------
-------------------+++
Chuck Ludwigsen <clud...@hotmail.com>
Sr. Engineer - Summit Data Group <chu...@summitdata.com>
President - Memphis Informix Usr Grp <mi...@hotmail.com>
+++-------------------------------------------------------------------------
-------------------+++
Manoj <ma...@mailhost.twowaytv.co.uk> wrote in article
<5vbc7r$9...@cssun.mathcs.emory.edu>...
>
> All,
>
> Are there any guidelines for the below ?
>
> ixda-RA idx-RA da-RA RA-pgsused lchwaits
> 4398 29 27805 32188 55
>
The Informix Online Performance Tuning course recommends Read Ahead
Utilisation should be 95% or greater.
--
David Williams
4 - (3 + 2 + 1) = < 2% & > 0.5% of 4
The explanation....
Add ixda-RA + idx-RA + da-RA (4398 + 29 + 27805 = 32232)
Subtract the result from RA-pgsused (32188 - 32232 = a negative number)
Since the result is a negative number you need to increase the RA
configuration.
You want the result of the calculation to be between 0.5% and 2% of
RA-pgsused. If the result of the calculation is above 2% you need reduce
the RA configuration.
Cheers, Tony.
#include <standard disclaimers>
----------
From: Manoj
To: 'inform...@rmy.emory.edu'
Subject: lchwaits, read aheads
Date: Friday, 12 September 1997 23:27
All,
Are there any guidelines for the below ?
ixda-RA idx-RA da-RA RA-pgsused lchwaits
4398 29 27805 32188 55
Thanks,
The correct formula is....
(1 + 2 + 3) - 4 = < 2% & > 0.5% of (1 + 2 + 3)
1) Add ixda-RA + idx-RA + da-RA (4398 + 29 + 27805 = 32232)
2) Subtract RA-pgsused from the result
32232 - 32188 = 44 = 0.14% of 32232
The result is less than 0.5%, so the advice of increasing the RA
configuration at least was correct.
Cheers, Tony.
----------
From: Rick Bernstein
To: tony edwards
Subject: RE: lchwaits, read aheads
Date: Monday, 15 September 1997 09:12
Won't your formula always produce a negative number?
I.E. Read-ahead pages used will be less than Read-ahead pages read.
==== ====
Perhaps the formula should read:
(1 + 2 + 3) - 4 = < 2% & > 0.5% of 4