Hello,

We use BPM processes extensively and we found an efficiency problem when number of instances increased.

We have transition documents published on broker and that is our problem. When number of transition documents is big we have a big queue on broker and each pass between process steps waits whole queue which takes a lot of time. Time necessary to finish a bpm process is a sum of processing transition documents (our issue) plus time required to configure systems.

Is there any configuration we could use to avoid this problem?

You should use OptimizeLocally QoS for all process when possible, unless you are clustered.

Review the use of logical servers, when used forces PE to publish a transition document.

I heard in wM 8 there is an option to priorize JMS, and use clustered brokers.

Hope this helps.