So, now that I have a firm grasp of what’s good and bad about WCF -> MSMQ, I’m going to take the evening out to review nServciceBus. It’s website isn’t down, and I have a pot of tea and a stack of old CD’s, so lets see how much I can figure out. The inherent pub/sub async distributed, queued architecture promises to be easier to configure and more fault tolerant than WCF. This should be a low bar…
And besides, there’s a whopping 35 entries under the nServiceBus tag on ServerFault, so how niche can it be? 😉
As a side note, I’m listening to Grumpy’s album “No Respect For Gravity”. Indeed.
How did your NServiceBus exploration go?
I very quickly got the pub-sub example going, and it worked really well. I spent only a few minutes trying to extend it to a multi-machine environment and wasn’t able to immediately grok how. I have a strong sense that I’m going to try to use this on my next distributed app instead of a monolithic broker style architecture.
Moving to a multi-machine environment entails changing the UnicastBusConfig section of the subscriber’s config so that its MessageEndpointMappings entry has MyPublisherInputQueue@MyPublisherMachine.
Hope that helps.