All, I need some design guidance.
We are implementing a file-transfer system that transfers files from agent software installed on customer machines into a central network (or to other agents in P2P fashion) over ActiveMQ. The files are transferred as binary-chunks [32K size]. We want to isolate each individual data-transfer using a dedicated queue. The dedicated queue will be created before the start of the transfer and dropped at the end of the transfer. The reason why we chose a dynamic multi-queue architecture is because 1) We dont have to apply any filters to identify messages intended for a particular agents. Because the size of the files could range from a few MB to a few GB, we foresee performance problems filtering a large number of messages for a large number of agents. 2) It provides for more isolation of the data of various customers instead of having them all in one or more predefined queues. The questions I have are 1) Is this an agreeable JMS practice? Since JMS API does not offer any queue management out of the box. 2) Is this agreeable ActiveMQ practice too? 3) Will there be any performance problems in storage and management of so many queues? 4) What other kinds of issues should I expect? 5) Any alternatives that can be suggested for this design? Many thanks for your kind suggestions. -- View this message in context: http://old.nabble.com/Need-design-inputs-on-programmatic-management-of-large-number-of-queues-tp26126813p26126813.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
