Hi Glen, > > I've been thinking about this and it seems to me there are > > two levels of > > mediation Synapse can offer: a low level model where there > > are messages > > flowing thru and u mess with them as they flow thru and a higher level > > model where Synapse virtually hosts services which it > > mediates forward. > > That means in those cases Synapse handles ?wsdl, ?policy etc. > > properly. > > (That's basically the WS gateway model).
Can we call these models message mediation and service mediation? > Right. I think the latter model is a good deal more interesting, and I > believe those are the kinds of use cases that people in real enterprise > environments are concerned with. I don't typically want something that > intercepts *all* my messages, I want a model for deploying individual > "virtual services" that I can manage, configure, and control > individually - and that can do things exactly as you mention with ?wsdl > ?policy etc.... So .. I've been there before. I'm the one who originally created the IBM Web services gateway which Paul took and made real :). It did *exactly* what you described and more. > Personally I think Synapse should end up focusing on the "service-aware" > model, with the option of "wildcard services" that you can configure as > catch-alls (or catch-alls for particular types of messages, even). This > kind of world would, I think, make it vastly easier to change the > configuration of per-service mediations without getting lost in a tangle > of non-relevant rules. (and yes, I realize that you can build this with > either model "on top" of the other... :)) We started on Synapse with a focus on message mediation. I'd like to get thru 1.0 with that focus and then look at service mediation. Why? I'm not convinced both can be done with exactly the same underlying processing model. It may be that service mediation can be layered cleanly on top of message mediation, but I'd rather not try to solve both problems at once right now. Once we have message mediation licked and delivered, let's start on service mediation. I'd rather see us deliver stuff incrementally rather than increase scope and get more muddied :). I think we have message mediation nearly under control but the processing model and config model are not quite rock solid yet IMO. I've been working on a cleanup with Asankha but haven't been able to put much time to it .. if all goes well we'll put a first cut tomorrow. (Basically we're reviewing the config model and processing model and trying to orthogonalize the design as needed.) Sanjiva. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
