Has anyone ever had any issues regarding transaction sequences in a clustered environment.
One example could be where you have a material master record being changed a couple of times within a split second in the material master application. You can see that it would be imperative that the order be preserved here.
My question is if the 2 changes are published to the Broker and you have mulit IS Servers processing these documents, how can you ensure the documents are processed in the correct order withOUT having to develop additional logic in your services.
I have already come up with a design to over come this problem, BUT it seems to hard and complicated to overcome this problem. Surely there must be an easier way by simply configuring webMethods to do this without the need for additional coding?
You can find information here about clusters and sequence number ideas. Taking sequential data and pushing out to a parallel handler will need to have some custom work done as the grain and ordering layers will be unique. (sequence every second document or process orignals before copies and after approvals)
Sorry there the “here” is an address I missed giving:
I found the link most useful. I also just wanted to make sure that this functionality must be coded…that is…its not provide by webMethods as part of the application.
I am not sure that this is built-in between the Broker and IS. To think about, you would have to have an internal sequence number given out with the document and checked when the document is returned. So if two people check out a document and return it in reverse order, who’s changes should take priority? If could be how the document left or it could be or it could be a ranking of the computer systems doing the work or a rank of the user logged into the computer. That would be a reason to do custom work on this. It would be nice to have sequencing like that out of the box, but it would require a manager application that is linked to the running servers, processes, and brokers handling those documents.