| Author |
Message
|
| skytorch |
 Posted: Fri Jun 14, 2002 6:45 am Post subject: How to achieve "Hot" standby QM |
 |
|
Apprentice
Joined: 10 Jun 2002
Posts: 47
Location: New York City
|
Hi,
I wonder what's a good way to achieve this goal: In a layout with a active QM and a standby QM, if active QM fails, I'd like to have standby QM to (a) pick up all persistent messages from active QM and (b) reduce system down time as much as possible, i.e. high availability and as transparent as possible to end users (may be server bound or client bound).
If I use shared disk between these two QMs, the standby QM need to be started which may take some time.
What's your suggestion ?
Thanks in advance.
Sky |
|
| Back to top |
|
 |
| mrlinux |
| Posted: Fri Jun 14, 2002 7:30 am Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
Well the best way is to have MQSeries clustering enabled. If you try to use a high availability option you can have several minutes or more with no qmgr active.
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
| skytorch |
| Posted: Fri Jun 14, 2002 7:49 am Post subject: |
|
|
Apprentice
Joined: 10 Jun 2002
Posts: 47
Location: New York City
|
Thanks Jeff.
How does it address the (a) requirement ? i.e. By using clustering, how does it have the standby QM pickup persistent messages of the active QM ?
Also, I'd like to have standby QM a real "standby" QM, i.e. all incoming messages go to active QM unless it fails. How do I customize QM clustering to achieve this ?
Thanks.
Sky |
|
| Back to top |
|
|
| mrlinux |
| Posted: Fri Jun 14, 2002 8:32 am Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
Well you wont have a standby QM they will both be processing messages,
The only way I know to achive your goal is to use OS Specific clustering,
which will generate an outage of 3 minutes or more
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
| skytorch |
| Posted: Fri Jun 14, 2002 9:02 am Post subject: |
|
|
Apprentice
Joined: 10 Jun 2002
Posts: 47
Location: New York City
|
Thanks , Jeff.
If there's no other options, I have to reduce this outage time as much as possible. I'm thinking of using SAN on Solaris to let standby QM share persist queue files of active QM. SAN is doing the replication in real time. I have one persistent application queue to replicate.
It seems to me that start a QM is a fairly quick process (with limited persistent messages) which in order of 10 seconds.
So I wonder what's break down of the time for standby QM to start ?
Therefore, how to optimize it to reduce system down time ?
Thanks.
Sky |
|
| Back to top |
|
|
| mrlinux |
| Posted: Fri Jun 14, 2002 10:23 am Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
Well the other issue you will have is the channels if you are using persistant messages, the channels maintain a sequence number between them to know if the other has received all of the messages sent.
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
|
|
