| Author |
Message
|
| jack_the_ripper |
 Posted: Mon Oct 06, 2003 12:28 pm Post subject: runtime db disaster recovery |
 |
|
Apprentice
Joined: 12 Feb 2003
Posts: 46
Location: NJ
|
Hi,
We are trying to put in a plan to recover workflow (3tier) when the runtime database goes down. There are two databases, one is runtime database and another is replication of this. The two databases (RT DB and its replication) are in two different sites and replication is sync with RT DB with the maximum of 15 minutes of delay. Ie in the worst case, there is 15 min data loss. We are wondering how we should change the workflow configuration details (for example, Runtime database properties in fmcrc on AIX) so that workflow uses a different runtime DB (ie replicated DB) when the original RT DB crashes. We apprecite any suggestions, thoughts, pointers on this. Please let me know if anybody implemented Disaster recovery for RT database in Workflow.
Thanx,
Jack0.
_________________
Jack,
IBM Certified in Websphere MQSeries
IBM Certified in Websphere MQSI
IBM Certified in Websphere MQWorkflow |
|
| Back to top |
|
 |
| Ratan |
| Posted: Mon Oct 06, 2003 12:40 pm Post subject: |
|
|
Grand Master
Joined: 18 Jul 2002
Posts: 1245
|
I dont think it is possible to dynamically change the location of Runtime DB in MQ Workflow. Is there any way?
You need to use OS level clustering (HACMP for AIX, Windows Cluster Server for Win2000) to acheive a failover for your Database server.
All the DB servers in the cluster should be writing / reading to/from the same physical media, so there wont be any latency in the event when one fails and other takes over. Also you can have the data mirrored, in case your physical media fails.
_________________
-Ratan |
|
| Back to top |
|
|
| vennela |
| Posted: Mon Oct 06, 2003 12:50 pm Post subject: |
|
|
Jedi Knight
Joined: 11 Aug 2002
Posts: 4055
Location: Hyderabad, India
|
Is there any reason why you are not using HACMP for databases (ofcourse other than cost)
As far as I know, fmcrc has the DB name but not the hostname where the DB is hosted. On the workflow server box, you will have DB2 client and catalog the DB2 instance (you give the hostame of the DB2 server and port number while cataloging). So, it is more of a DB2 question rather than a workflow question. |
|
| Back to top |
|
|
| cowboy2003 |
| Posted: Fri Feb 27, 2004 10:45 am Post subject: it should work |
|
|
Novice
Joined: 27 Feb 2004
Posts: 22
|
Here are three way to failover RT db
1: os cluster with db2 partition, such as HACMP on AIx and MSMC on windows. It is best solution, however it is expenseive.
2: db2 RT Db duplication: using db2 duplication server. it is reconneded for RT db.
3 Db2 log ship. I believe you are using this solution. Here are two possibel solution
3.1 change ODBC configuration, point to backup machine.
or
3.2 change the backup machine ip and name to primary one.
Hope it help. |
|
| Back to top |
|
|
| Ratan |
| Posted: Fri Feb 27, 2004 10:56 am Post subject: |
|
|
Grand Master
Joined: 18 Jul 2002
Posts: 1245
|
only solution (1) is Hot Failover.
I dont think (2) and (3) are Hot Failovers, or may be I did not understand ( I am not good at databases).
_________________
-Ratan |
|
| Back to top |
|
|
| cowboy2003 |
| Posted: Sat Feb 28, 2004 3:29 pm Post subject: |
|
|
Novice
Joined: 27 Feb 2004
Posts: 22
|
You are right. It is hot failover. No all user SLA requires hot failover.
_________________
IBM Certified Workflow Solution Expert |
|
| Back to top |
|
|
|
|
