The correlation Id is useful when a reply queue is shared by processes or by producers/consumers concurrently and the replies come back out of order. Typically, a queue is used to decouple 'requests for work' from 'work' so the order of 'work' completion is not deterministic, hence the need for a correlationId. If all requests are serialized as in an RPC style interaction there is no need for the correlation id, but then why use a MOM at all.
There is some more detail at http://activemq.apache.org/how-should-i-implement-request-response-with-jms.html On 8 July 2010 11:32, sic <sic_1...@naver.com> wrote: > > I give thought to the correlation ID concept on the JMS Specification. > > When using synchronous queue messages, we should have JMSReplyTo header > which has a destination of the producer. > > After sending, the producer waits for reply message to reach its temporary > queue. > > In this format, it is only the consumer of that message that can send the > reply message by JMSReplyTo header. > > So it's fine without correlation ID because the producer waits for reply > message not on the static queue but on the temporary queue registered in the > JMSReplyTo header. > > And that's all in my thought. > > Then, What's the purposes for using the correlation ID? > > I have heard that using this is for relating sending message to its reply > message. > > But it doesn't need to relate between them unless we using it in the > application layer. > > Although it's not about activemq, I try asking because I can't find any > appropriate forum for that issue. > > I hope anybody informs me of this concept. > > regards, > > sic > -- > View this message in context: > http://old.nabble.com/correlation-ID-on-synchronous-message-tp29105526p29105526.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. > > -- http://blog.garytully.com Open Source Integration http://fusesource.com
