CODEWITHSUNDEEP
ibm-mq
M.I
M.I

Reputation: 303

Intended use of Transmission Queue

This is very basic question about IBM WebSphere MQ V7.

Regarding the Transmission Queue, my understanding is it is only used with remote queue that resides in the same queue manager. Therefore, if I want to put message to the queue, I need to put it to remote queue.

It is like this.
App --> Remote queue --> Transmission Queue

My question is:
Is it possible to put the message directly into transmission queue like this?
App --> Transmission Queue

--Modified on 2014.03.17 --

I found a way to put message directly into transmission queue. I do not know this is a common use, but in order to do that I needed to prepend MQXQH to the message. I tried and confirmed it works. See the Infocenter reference here.

Upvotes: 3

Views: 3426

Answers (3)

T.Rob
T.Rob

Reputation: 31832

Although it is possible to put messages directly on the XMitQ, there is considerable risk in allowing that to occur so most admins will prevent applications from directly accessing that queue. As you have found, it is possible to construct a message with the transmission queue header and behind that a normal message with the MQMD and payload. (This is, in fact, excatly how the MCA works.)

The problem here is that the QMgr does not check the values in the MQMD residing in the payload so you can put mqm as the MQMD.UserID and then address the message to the remote command queue and grant yourself admin access to that remote QMgr.

Security-conscious administrators typically use two security controls to prevent this. First, they disallow direct access to the XMitQ. That helps for outbound messages. More importantly, they set the MCAUSER of their RCVR/RQSTR/CLUSRCVR channels to a non-admin user ID that is not authorized to put messages onto any sensitive queues.

The other issue is, of course, that what you describe completely defeats WMQ's name resolution. By embedding routing into the app, you prevent the administrator from adjusting channel weights, cluster settings, failover and load distribution at the network level. Need to redistribute traffic? Redeploy the code. Not a good plan.

So for security reasons and because you paid a lot of money to get WMQ's reliability - much of which comes from dynamic addressing and name resolution features - coding apps to write directly to the XMitQ is strongly discouraged.

Upvotes: 3

Chris Tarpy
Chris Tarpy

Reputation: 70

You should not directly be using the transmission queue. Its used by the message channel agent (MCA) as temporary storage when sending messages across to a remote queue manager.

This is distributed queuing - i.e you publish a message to Queue Manager A, and want it routed to a local queue on Queue Manager B. So you define a reference on QM-A referring to the local queue on QM-B. This reference is the 'remote queue definition'.

The remote queue definition specifies the transmission queue name. The transmission queue is bound to the MCA, which in turn knows about the remote QM.

Upvotes: 1

Roger
Roger

Reputation: 7476

Do not ever put directly to a transmission queue. It is dangerous if you do not know what you are doing.

You should put your message to a remote queue. A remote queue is not the same as a local queue. A remote queue is simply a pointer to a queue on another queue manager.

Upvotes: 3

Related Questions