Hi,
I have a real time data services job which is doing the following
a) read and un-nest a MQ message
b) perform basic validation using lookups to a mainframe DB2 table
c) Load the validated data to another Mainframe DB2 table
There are 2 input streams to this job, the first involves users ‘publishing’ a single page worth of catalogue data which would involve up to 20 messages being sent. The second stream occurs if a user ‘publishes’ a full catalogue worth of data, usually at the end of a day. This could contain 2000+ messages and multiple catalogues could in theory be published on the same day.
We have been taking timings based on a catalogue which publishes 1000 messages and these are taking 30 mins to process.
We have played around with various setting for DoP, max instances, min instances etc without making a lot of difference to the timings. A saving of 3 mins is the best.
By monitoring the jobs has they run, we believe that the bottleneck is in reading the message queues. Has anybody any suggestions as to how we could speed this process up.
Thanks
PeterElliott 
(BOB member since 2014-02-13)