[siakadoumbia]

Hi All,

I need help to clarify a situation we are facing on our production environment.
We are running Sybase on an IBM server :

> lsattr -El sys0

SW_dist_intr false Enable SW distribution of interrupts True
autorestart true Automatically REBOOT system after a crash True
boottype disk N/A False
conslogin enable System Console Login False
cpuguard enable CPU Guard True
frequency 484000000 System Bus Frequency False
fullcore false Enable full CORE dump True
fwversion IBM,RG050405_d79e05_r Firmware version and revision levels False
id_to_partition 0X036FC37434C8B201 Partition ID False
id_to_system 0X036FC37434C8B201 System ID False
iostat false Continuously maintain DISK I/O history True
keylock normal State of system keylock at boot time False
maxbuf 20 Maximum number of pages in block I/O BUFFER CACHE True
maxmbuf 0 Maximum Kbytes of real memory allowed for MBUFS True
maxpout 33 HIGH water mark for pending write I/Os per file True
maxuproc 1024 Maximum number of PROCESSES allowed per user True
minpout 24 LOW water mark for pending write I/Os per file True
modelname IBM,7038-6M2 Machine name False
ncargs 6 ARG/ENV list size in 4K byte blocks True
pre430core false Use pre-430 style CORE dump True
pre520tune disable Pre-520 tuning compatibility mode True
realmem 33554432 Amount of usable physical memory in Kbytes False
rtasversion 1 Open Firmware RTAS version False
systemid IBM,01652256A Hardware system identifier False

> lsattr -El aio0

autoconfig available STATE to be configured at system restart True
fastpath enable State of fast path True
kprocprio 39 Server PRIORITY True
maxreqs 4096 Maximum number of REQUESTS True
maxservers 10 MAXIMUM number of servers per cpu True
minservers 1 MINIMUM number of servers True


We are currently facing a performance issue with the database, which I think to be related to the that Sybase has been running for a very long time :

> ps -ef | grep dataserver

sybase 548944 7802912 120 Jul 22 - 57243:25 /apps/sybase/12.5/ASE-12_5/bin/dataserver -ONLINE:2,0,0xffffffff90004044, 160000000, 0x80000000, 0x1159
sybase 1126536 7802912 29 Jul 22 - 56980:35 /apps/sybase/12.5/ASE-12_5/bin/dataserver -ONLINE:6,0,0xffffffff90004044, 160000000, 0x80000000, 0x1159
sybase 2547888 7802912 26 Jul 22 - 57586:32 /apps/sybase/12.5/ASE-12_5/bin/dataserver -ONLINE:3,0,0xffffffff90004044, 160000000, 0x80000000, 0x1159
sybase 4669454 7802912 56 Jul 22 - 58077:53 /apps/sybase/12.5/ASE-12_5/bin/dataserver -ONLINE:4,0,0xffffffff90004044, 160000000, 0x80000000, 0x1159
sybase 6488244 7802912 41 Jul 22 - 57935:43 /apps/sybase/12.5/ASE-12_5/bin/dataserver -ONLINE:1,0,0xffffffff90004044, 160000000, 0x80000000, 0x1159
sybase 7540944 7802912 104 Jul 22 - 57219:53 /apps/sybase/12.5/ASE-12_5/bin/dataserver -ONLINE:5,0,0xffffffff90004044, 160000000, 0x80000000, 0x1159
sybase 7802912 749568 120 Jul 22 - 62154:42 /apps/sybase/12.5/ASE-12_5/bin/dataserver -sKOR_FR_XXX_ASE -d/dev/rlv_SKORP_MAS -e/apps/sybase/adm/KOR_FR_XXX_ASE/log/errorlog_KOR_FR_XXX_ASE.log -c/apps/sybase/adm/KOR_FR_XXX_ASE/cfg/KOR_FR_XXX_ASE.cfg -M/apps/sybase/adm/KOR_FR_XXX_ASE/cfg


Now the sysadmin wants to reconfigure to values minservers=10 and maxservers=20.
But I understood from a previous post (Fasherman) that databases accessing raw devices get little benefit of aioservers configuration, and our database uses exclusively raw devices (except tempdb).
What do you think about this matter ?
Your help will be very mauch appreciated.

Regards,
Siakman

[lawrenzo]

what kind of performance issues are you facing?

We have just been through about 6mnths of performance issues and after analysing nmon data over a two year period the evidence pointed to full system utilization.

We upgraded the server and performance improved considerably:

check out the belo thread when I was investigation AIO

http://www.pseriestech.org/forum/aix...onous-133.html.

Other things to do is run scripts to collect system info ie we run a ps awux command as deamon that logs to a file every 15mins.

do you have nmon - check things like memory, cpu, paging etc utilization per day, 95 percentile and also top process's hammering the box.

you may find that on 21 days ago a new process / script / install has occured and this is actually causing poor performance.

To answer you question:

I studied VIO and the performance and tuning guide does state that no real issues can be made from using AIO on raw database chunks however not sure if this is gospel. For oracle on mounted filesystems the formula's stated in the link help the system no end.

First port of call is to determine what is using all cpu.

Hope this helps man

Chris.