| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
| Performance questions |
« View previous topic :: View next topic » |
| Author |
Message
|
| ussm120 |
 Posted: Mon Nov 05, 2001 12:53 am Post subject: |
 |
|
Newbie
Joined: 28 Oct 2001
Posts: 9
|
Hi,
we test MQ performance on the following infrastructure:
Smal RS6K, a 16MB LAN, 2MB WAN and A big RS6K. the resluts are:
- Production of messages
The smal RS6K initiates the sending and is to produce up to 200+ messages per second
Maximum number Processed messages:
100 character messages(115 msg/s)
1K character messages(70 msg/s)
4K character messages(25 msg/s)
Maximum real available bandwidth
800 Kbit/sec
WITH CONTROLLED QUEUING
100 character messages(60 msg/s)
1K character messages(45 msg/s)
4K character messages(20 msge/s)
Average real available bandwidth
600 Kbit/sec
Greetings,
Mohammed
|
|
| Back to top |
|
 |
| dschnaider |
| Posted: Thu Feb 07, 2002 1:30 am Post subject: |
|
|
Newbie
Joined: 06 Feb 2002
Posts: 5
|
I already have a HP machine with 8 CPU's, 8GB of memory, fibre channel... etc...
and I have a poor performance of MQSeries.
If you have any tip to improve performance, please let me know.
It can be on the os, admin or programming.
One problem that we find out is that when you put 10 messages and then commit (it can be any number of messages, let say n) MQseries will make N*2 IO's to the disk. When first they write each header (just 8 bytes long) and then the body (in my case 4KB). The fibre channel uses blocks of 16KB... It is a big performance problem.
_________________
Daniel Schnaider
dschnaider@yahoo.com |
|
| Back to top |
|
|
| oz1ccg |
| Posted: Tue Feb 12, 2002 1:35 am Post subject: |
|
|
Yatiri
Joined: 10 Feb 2002
Posts: 628
Location: Denmark
|
I've got some information to compare some of the platforms from the latest MQSeries Conference (Vienna 2001) supplied by IBM, and they show these figures:
Non_pers Get/put Persistent Get/put/Cmit
MQSeries Version 5.1 5.2 5.1 5.2
AIX RS6000 43P-260(2CPU) 5000 9200 100 1200
AIX RS6000 S70 (12 CPU) 33000 80000 100 2200
NT Netfinity 7000M10(4 CPU) 14000 22000 100 900
HP_UX K380 (4CPU) 6000 10500 65 800
Solaris Enterprise Ultra 450(4CPU) 8000 12000 80 1200
Setup of the test application:
- Trusted application binding
- One application per processor
- Persistent: many(50) parallel applications
- 1k messages
- Seperate queues per applications
So the result is, get up and run on version 5.2 asap if you got performace problems.
_________________
Best regards
Jørgen Pedersen
IBM Certified MQSeries Specialist
IBM Certified MQSeries Solutions Expert
[ This Message was edited by: oz1ccg on 2002-02-12 01:39 ] |
|
| Back to top |
|
|
| mrlinux |
| Posted: Thu Feb 28, 2002 12:35 pm Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
V5.1 of MQSeries on HPUX was a dog, The release of v5.2 improved the performance dramaticly. My converstations with IBM was that it led to a 20 times
improvement in speed.
Well the K-Class machines have 200mhz RISC processors on them, so trying to compare that to a 2GHZ processor is slightly unfair.
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
