E
Edmond wong
Hi,
I have a design question when we try to interface our multiple database
systems with other party using Weblogic. The problem is as followed:
1. We listen to a request message on startup class from the other party
to send a query to 3 of our databases. There is no dependency to
retreive the data among the databases.
2. To achieve concurrence, we split the request message into 3 messages
to JMS queue, and have MDB instances to process the queue and send the
request to 3 database systems.
3. From here, we got a result set in each instance of the MDBs. We would
consolidate the result sets somehow and send a single response to the
other party.
The question is what is the best way to consolidate the result sets from
the MDB instances. There are also the following requirement we have to
consider:
1. We want to use two clustered weblogic servers for load balancing, and
fail over. So, we cannot call a singleton from startup class to
consolidate the results.
2. We don't want to use the database to store the results because of
potential performance issues.
I wonder if there is a magical way, or some sort of memory space shared
between two clustered weblogic servers for the purpose. If so, can you
help to point me to the right direction. Any other suggestion is welcome.
Thanks,
Edmond
I have a design question when we try to interface our multiple database
systems with other party using Weblogic. The problem is as followed:
1. We listen to a request message on startup class from the other party
to send a query to 3 of our databases. There is no dependency to
retreive the data among the databases.
2. To achieve concurrence, we split the request message into 3 messages
to JMS queue, and have MDB instances to process the queue and send the
request to 3 database systems.
3. From here, we got a result set in each instance of the MDBs. We would
consolidate the result sets somehow and send a single response to the
other party.
The question is what is the best way to consolidate the result sets from
the MDB instances. There are also the following requirement we have to
consider:
1. We want to use two clustered weblogic servers for load balancing, and
fail over. So, we cannot call a singleton from startup class to
consolidate the results.
2. We don't want to use the database to store the results because of
potential performance issues.
I wonder if there is a magical way, or some sort of memory space shared
between two clustered weblogic servers for the purpose. If so, can you
help to point me to the right direction. Any other suggestion is welcome.
Thanks,
Edmond