Soliciting Biztalk BAM scenarios and issues

How great it would be if every features we put in to the software is what customers need/want.   Anyway, I am creating this entry to solicit your input on how you use or plan to use BizTalk BAM.   Please response as a comment with your scnenario.

Do also tell us what feature you like and dislike.



Comments (5)

  1. Eric Stott says:

    MES and OES are great, but what would be even better would be to have a datastore when Orchestrations do fail, that the ‘up till it failed’ BAM data would be stored. This would help, as then it would be a central repository for the support team to go to, and it would also not pollute the live BAM data. Call it Failed Messaging Event Stream FMES and Failed Orchestration Event Stream FOES, or better yet, just have it in MES and OES automatically.

  2. keithlim says:

    The failed info is currently available in Biztalk HAT (Healt Activity Tracking), but if you want to have it in BAM, the DirectEventStream and BufferedEventStream could be used; this would allow you to persist BAM events regardless/independent of the transaction of pipeline or orchestration.

    Some interesting scnenario in this usage could be that, you use BES.BeginActivity but then something goes wrong in the pipeline processing and the remaining  BES.UpdateActivity (such as Milestone_Processed_Successful) and BES.EndActivity do not get processed.  You can set up an alert based on if there the duration of processing (time from received till Process Sucessful) is greater than certain time, send me an email, or that if it does not get processed at all, send me an email. You can create all kinds of alerting to tell you when something goes wrong.

    Alerting can also be used to monitor aggregation, e.g. if the average process time is greater than certain number, alert me, then you can use this to look into where is the bottleneck.

    I will blog on alerting later on… this is a new feature in 2006.

  3. Eric Stott says:

    However, the purpose in BAM is to be able to see ‘trends,’ in your business data. I know that you could use DES or BES, but that is having to manually code for errors, whereas if it were automatically encoded in OES and MES, it would add so much more value. Along with the alerts that are available in 2k6, if it had its own datastore, then support could go to it and find patterns in offending data, and then be able to go back to the developers and create solutions to the bad data, or be able to go back to the source with a historical record.

  4. paul wu @ says:

    I am on a project where a BAM activity is initiated in a receive pipeline, add the incoming message as a related reference, tracked the milestones in orchestrations, and end the activity in the send pipeline where we also store a copy of the outgoing message as a related reference. We then have a custom web app to browse the bam data, review orders received, orders sent, and resubmit or redeliver orders. Debugging is tricky with multiple continuation. Can use more documentation on the subject.