This have been a long thread... and I'm going to extend it a bit more..... ;o)
I like to point some things out with regard to the security, and I agree with Paul and T.Rob to some extend...
Without security exits and/or SSL there are no authentication at all. I doesn't matter if you use SDR/RCVR or SVR/RQSTR or the other combinations including cluster channels. If we're using a RQSTR then it uses callback if the RQSTR initiates the conversation, but what happens if it was the sender, does it break the connection ?? NOPE, it just acts as a normal RECEIVER channel. What about a fully qualified SVR channel, if this one starts the connection no callback, just a normal SDR channel. And the Cluster channels, acts allmost the same way, if you can connect to the full-repos qmgr, you have access to the whole cluster (if there are no security exits/SSL on the CLUSRCVR channel).
My personal advis is use SDR/RQSTR because you (normally) can start the channel from both ends, and I don't care what the partner in the other end says I allways use RQSTR with some exits.
Currently I'm working on a security authentication exit for z/OS and the Distributed platform. And the technic behind it seems to work. The challange is to archive a reasonable security with a low cost, no problem with high security just use SSL or a 3 party product.
Just my $0.02 ;o)
Kind regards Jxrgen
www.MrMQ.dk
T.Rob,
I agree entirely with you that you should employ some form of mutual authentication such as SSL. However, I do not see this as just a Requester/Server issue but of any channel pairing. I had assumed, perhaps wrongly, that a channel was always protected in this fashion if the data was of a sensitive nature. I would not like people to get the impression that requester/server pairs are somehow inherently unsafe and that sender/receiver pairs are inherently safe. The reality is that all channels should employ something like SSL if one needs to be certain of the authenticity of the data and its source.
I also agree that a callback is not an excessive overhead in the grand scheme of things. However, it is still overhead and extra complexity. If users feel safer by having requester/sender pairings then that's fine by me. I was merely trying to point out an alternative.
Cheers, P.
Paul G Clarke WebSphere MQ Development IBM Hursley
_________________________________________________________________ Fe alle de nye og sjove ikoner med MSN Messenger http://messenger.msn.dk
Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive
