-
How to add clustering for message processing that need sequencin (2 messages)
- Posted by: Satish Babu
- Posted on: December 15 2006 20:46 EST
I have messages in a MQ queue that needs to be processed in the same sequence as it was received. The solution needs to be highly available. I can not use a MDB since it does not guarantee sequencing. A simple stand alone Java program running on one single machine will not provide failover capabilities. What solution can I use to process the message in a sequence and still provide the capability of automatic failover. We have option of using Websphere 6.x or an ESB for this solution. Thanks SatishThreaded Messages (2)
- Re: How to add clustering for message processing that need seque by Nati Shalom on December 16 2006 23:33 EST
- Re: How to add clustering for message processing that need seque by Christopher Stach II on December 18 2006 09:43 EST
-
Re: How to add clustering for message processing that need seque[ Go to top ]
- Posted by: Nati Shalom
- Posted on: December 16 2006 23:33 EST
- in response to Satish Babu
Hi SatishI have messages in a MQ queue that needs to be processed in the same sequence as it was received. The solution needs to be highly available... What solution can I use to process the message in a sequence and still provide the capability of automatic failover.
We have been involved in similar scenarios in the past specifically with MQ. One of them was with one of the largest exchanges organisation. The requirements was to take messages from MQ which was delivered by another part of the organization parse them, validate them and store them in an oracle 10G data warehouse for further processing. We used our space based messaging approach combined with master/worker pattern to achieve that. The master/worker on spaces are transactional and ensure that a message that that was retrieved from MQ and stored in the space will not be lost until it was successfully processed. Processed in this case means stored in the oracle 10G. The messages can be retrieved in the same order they were written. With this approach we were able to push 300-400 million rows processed in 1:45 hours using 2x4-Way Itanium processors. You can find an interesting presentation that was given to Gartner by one of the group members of this organisation. HTH Nati S GigaSpaces Write Once Scale Anywhere -
Re: How to add clustering for message processing that need seque[ Go to top ]
- Posted by: Christopher Stach II
- Posted on: December 18 2006 09:43 EST
- in response to Satish Babu
ActiveMQ does/attempts to guarantee message sequence across a cluster.