MQSeries.net :: View topic - Data driven flows in Websphere Message Broker (version 6)

jstorni · Posted: Wed Jul 12, 2006 6:50 am Post subject:

Hi all,
We have to implement 400+ use cases for request/response to external systems. The use cases are splitted 60/40 in two external systems. Both requires transformations (XML->something). All the use cases related to a given external system, share the same message format (variable field list, with delimiters).
In addition to these use cases, we have to implement 200+ use cases of business logic.

Currently we have a component that creates the target message by accesing a repository that contains the message description of each use case. So, the only requirement for transformation is the creation of a XSLT file for field mapping definition only.

The questions:

1) Do you have experience in an implementation with Message Broker 6, with 600+ flows ?

2) We are analizing to create two generic "data driven" flows, that handle the requests by identifying the service in runtime and loading the required layout data to create both request or response messages. With this approach, we have to implement just 2 (but a bit complex) flows instead of 400+, and develop 800+ XSLT templates for field mapping. In addition, the XSLT templates can be automatically created (at least the skeleton), because we have a complete inventory of the input and output fields. What do you think ? Is this approach aligned with the common patterns used with IBM Message Broker ?

3) In relation to the previous: do you think if implementing the 400+ integration use cases with DataStage TX extender for IBM Message Broker 6, building manually the type tree and the field mapping both for request and response (800+ type tree and 800+ mapping definition), is a better solution ?

Many thanks !