Trading Networks running on multiple clusters

We are currently upgrading Webmethods from version 7 to 9. We are also implementing a rearchitecture of our server configuration during this upgrade. We currently have one unix cluster with 2 nodes, where our entire code base is stored on this cluster. Our code base is nearing 400 packages, which takes nearly half of our heap upon startup. In our rearchitecture, we plan to have 2 clusters with 2 nodes each, and we plan to split our code base as evenly as possible between the 2 clusters. We will have Trading Networks running on each cluster, processing transactions for that cluster’s code base. My question: With this new configuration, is it possible through Trading Networks to have one full, seamless view of all transactions, regardless of which cluster the tansaction was processed on? (ie If a transaction is processed in cluster 1, can I configure TN to see this transaction from cluster 2?) Thanks.

Really this is interesting question.

I don’t believe this can happen, but you can use 1 MWS to see all transaction just by use of drop down of box for servers in transaction anayalsis page

May be some one in forum can help you further

If both clusters are pointing to the same TN DB, then you will see all transactions going through both systems.
I can’t see any reason it won’t work.

Sorry i missed that point

Yes both Clusters pointing to same DB will work

Yes if you configure the TN to use the same DB and accounts and you can view same transactional info and it should be work in this topology :smiley:

Thanks for the replies. So if I have different clusters pointing to the same TN DB, how does the processing rule in the DB know in which cluster the code to execute lives? Remember that we will not deploy all code to all clusters. Each cluster will contain a (mostly) unique subset of our full code base. Thanks.

Good point made but If you choose to have the same DB then yes same code base should be in sync across all the clusters IS nodes and we didn’t focus your question on splitting your packages (400+) to the different IS’s due to heap load.


how do you plan to route the incoming transactions to different clusters?

Outbound should not be an issue.

You need to configure below property to make TN run in cluster environment