MQSeries.net :: View topic

Senthamizh · Posted: Fri Nov 04, 2011 6:28 am Post subject:

Hi

I read through some of the earlier posts on this forum regarding the comparison of performance between compute node and DB node. And what i inferred was that the performance in MB mainly depends on how the message tree is handled, since compute node copies the entire tree and allows to change the tree structure, it occupies more memory than the DB which doesn't allow tree structure changes.

Now, in my current project, we are using 4 compute nodes at various stages of the flow only to set some environment variables specific to that flow(like sourcesytem, targetsyetem, status, etc..) for logging purpose.

My doubt is that, if we replace the 4 compute nodes with 4 DB nodes(per flow) for setting these environment variables then does it increase the performance to a considerable amount or will it neglible.

Considering that we have the same kind of logging pattern for more than 300 flows, I believe that using DB nodes for environment mapping for the all flows would definetly have an considerable increase in performance of the overall broker.

Let me know if my consideration is correct.