I have a high transaction volume application that we are "modernizing" and have two camps on the best approach. The application firewall will be a DataPower appliance. The question comes from the use of BPM for the Orchestration layer vs. WMB for the Orchestration. All our basic services are WMB/MQ wrapped.
Does anyone have either some metrics to compare, or strong opinions on which approach will work the best. We are talking a million or so transactions per day.
Thanks in advance.
NOTICE: developerWorks Community will be offline May 29-30, 2015 while we upgrade to the latest version of IBM Connections. For more information, read our upgrade FAQ.
This topic has been locked.
2 replies Latest Post - 2013-03-21T04:25:25Z by SystemAdmin
Pinned topic Anyone have any high transaction volume metrics on BPM vs. WMB
Answered question This question has been answered.
Unanswered question This question has not been answered yet.
Updated on 2013-03-21T04:25:25Z at 2013-03-21T04:25:25Z by SystemAdmin
bmruter 110000S42H13 PostsACCEPTED ANSWER
Re: Anyone have any high transaction volume metrics on BPM vs. WMB2013-03-20T14:37:25Z in response to SystemAdminThis sounds like a conversation I have frequently. Both of these technologies will scale to what you need. Broker (at least in our situation yours may vary) is more cost effective at our shop (for many reasons I wont discuss here). The question comes down to the capabilities you need. If the process is long running, interuptable, requires human itervention, or needs transactional or compensation capabilities then WPS is usually the answer. If none of those capabilities are needed, then I would go with broker, although there are other non functionals to consider, like liscensing, infrastructure avaialbility, developer skills, etc. If the million transactions are spread throughout the workday then the volume is pretty small for either of these techonlogies (BPEL at least, not sure what BPMN would do with that). If you need to crank these through in a fairly large burst then you'll have some performance tuning to get this to work but still doable.