How are you transitioning code design pubsub from ES to IS

We haven’t transitioned from 5.0/4.6 to 6.1 but we are going to be in the next month or so. Coming from a standpoint that I’ve worked with Enterprise Server about 80% of the time, I’m curious as to how other people are designing solutions and services with respect to publish-subscribe nature.

Take for example an Enterprise Server scenario where you have to parse a file, build xml, and then do some custom processing. A typical solution would be that component #1 belongs to an IO Adapter, component #2 belongs to an XML Adapter, and for arguements sake, component #3 belongs to an ILA Adapter. The only way to complete this solution would be to perform a publish-subscribe from one component to the next.

Now taking that same scenario to Integration Server 6.1, there is in essence no absolute need to do a publish-subscribe. You could have a main servic that can call sub-services, thus acting no different than a sequential program.

So, my overall question is this: Have you moved away from publish-subscribe architecture and only use it when you need to send/receive data from separate Integration Servers? Or are you still doing the same amount of publish-subscribe, and if so, why?