Poor System Performance

[Former Member]

Hi All,

I am facing very big performance problem in QAS server. DEV and QAS in one server with 10gb ram.Hit

ratio in ST02 is more than 98% but the transaction takes more time to execute. My server AS/400 V5R4 running ECC5.0. One month back update HR support package, while updating error happens.then i clear the entries from tables and updated successfully. After that the performance gone down. This 75 gb for swap space is very high for this memory. This is my wrksyssts.

% CPU used . . . . . . . : 18.1 System ASP . . . . . . . : 864.4 G

% DB capability . . . . : 7.8 % system ASP used . . . : 37.5318

Elapsed time . . . . . . : 03:01:37 Total aux stg . . . . . : 987.9 G

Jobs in system . . . . . : 7509 Current unprotect used . : 40151 M

% perm addresses . . . . : .012 Maximum unprotect . . . : 43314 M

% temp addresses . . . . : .062

Sys Pool Reserved Max -

---

DB----- Non-DB- Act- Wait- Act-

Pool Size M Size M Act Fault Pages Fault Pages Wait Inel Inel

1 1839.91 233.50 +++++ .0 .1 .0 .1 87.9 .0 .0

2 7356.56 2.48 422 29.3 259.5 53.1 142.4 2325 .0 .0

3 484.03 .00 122 .0 .0 1.7 1.9 4.3 .0 .0

4 .25 .00 5 .0 .0 .0 .0 .0 .0 .0

This is my wrkdsksts values

Size % I/O Request Read Write Read Write %

Unit Type (M) Used Rqs Size (K) Rqs Rqs (K) (K) Busy

1 4328 123488 37.5 390.3 5.0 20.3 370.0 8.2 4.8 18

2 4328 123488 37.5 194.4 6.3 45.1 149.3 8.8 5.5 16

3 4328 123488 37.5 240.0 5.1 33.8 206.2 4.1 5.2 14

4 4328 123488 37.5 197.6 7.3 38.8 158.8 5.6 7.7 12

5 4328 123488 37.5 279.3 5.3 18.9 260.3 7.0 5.2 16

6 4328 123488 37.5 296.0 5.1 24.3 271.6 6.3 5.0 17

7 4328 123488 37.5 59.5 9.6 13.0 46.4 6.2 10.6 7

8 4328 123488 24.7 8.1 26.6 .0 8.1 .0 26.6 1

Thanks and regards

Shylendra

[Former Member]

Hi Shylendra,

your current swap is: 40151 M (not 75 GB)

Even when you say, that you dev system is good and only qas is bad, you should apply note 428855 and to get rid of pool 3 & 4 as well.

Then you should try and bump up pool 2 a bit - I waint until you did the rest and see the new results. at the moment you still seem to have page faults in pool 1

You should see that 40GB swap compared to 7.3GB Pool 2 is pretty small and not too good.

You should stop QAS and use DLTR3PKG.

If the problem stays the same, my guess is: DEV is used and QAS is rarely used and therefore "paged out" - you could then only split dev & qas into different pools which will make the performance worse for both systems.

As you are very memory constraint, you should reduce the buffers in ST02 as far as possible.

Regards

Volker Gueldenpfennig, consolut international ag

http://www.consolut.de - http://www.4soi.de - http://www.easymarketplace.de

[christian_bartels]

Hi Shylendra,

the overall system values that you copied do not look too bad - however, they are average values (you did not copy the elapsed time), so there could be higher values in shorter intervals.

Anyway, if there is a specific transaction that is showing poor performance, then the first approach should be STAD (or ST03) to find out if the problem is in the DB time, the CPU utilization or any other factor. If you have poor DB response times, then you could use an SQL trace (ST05) to identify the statement(s) that need a lot of time and analyze them for possible optimizations.

Kind regards,

Christian Bartels.