Data Queues vs. MQSeries
January 14, 2009 Timothy Prickett Morgan
Does anyone have any knowledge or references related to the difference in performance between doing a put to a journaled data queue vs. doing an open and a put (no connect) to an MQSeries queue?
In MQSeries, opens are expensive, so if you keep the queue open and do multiple puts, you will get very good performance. Using a data queue, you will get better performance as long as there are not too many entries in the data queue. If the data queue overflows (no telling where this might occur, but I have seen it around 100 messages), then both gets and puts to the data queue get very slow. This does not happen with MQSeries.
The question and answer given above is from the IT Jungle Web Forums. Feel free to add more to the discussion.
In V4R5, IBM added two parameters to the Create Data Queue (CRTDTAQ) command to help keep data queues as small as possible. The SIZE parameter allows you to specify the initial and maximum number of entries to be placed on a queue. The AUTORCL parameter allows the system to reclaim storage as needed. See Manage the storage used by a data queue for more information.