STORM-481 help

[Simon Cooper <si...@featurespace.co.uk>]

We're doing some performance testing on our topologies to try and increase the performance, and we've discovered that the section of the topology that is the current bottleneck needs to be parallelised. Unfortunately, STORM-481 is preventing us from parallelising that section and so increasing the performance. We've got some specific performance targets in our contracts that we are not meeting, so this is a serious problem for us.

I've tried to debug the issue, but I quickly got lost in the largely uncommented trident code, multiple thread communication, and the batch tuples being sent all around the topology.

Can anyone offer any tips or pointers as to what might be going wrong, or any specific areas I should be concentrating on?

Thanks,
SimonC