Handling Message Volume Spikes

[Greg Tataryn]

We are having issues with appearant spikes in volmes of messages and are researching ways to try to process faster or limit slowdowns on the engine. we have 7 IB threads feeding a single OB thread. 6 of the 7 are relatively low volume but the 7th is tens of thousands of messages per day and it is the one that seems to spike.

When we get an apparant spike, the OB queue starts to gradually back up more and more. There is no translation and no TCL performed on the messages. The IB does ACK the messages and the OB waits for an ACK but just kills the ACK when received. We have SMAT turned on for both the IB and OB. We have had some internal disucssions on multithreading but without using any translation we are not sure if that applies to us or not.

EDIT:

Sorry, should note that we have set up a “loopack” where we have the IB thread on one process with an OB that just sends back to the engine. The IB side of the loop is then on a spearate process with the OB thread to our HIS.

Topic