Re: FW: Reliable Messaging Summary

On Wednesday, March 5, 2003, at 01:17  PM, Walden Mathews wrote:

> This is a fascinating observation.  In other words, "reliability"
> can only be defined in terms readily understood within the
> application "domain of discourse"?  Or is there some other
> level?

I don't think I'd use the word "only". On the other hand, it's
plausible that "reliability" has quite different meanings for, say,
video-on-demand and banking. For some applications, predictable
jitter is more important than exactly-once semantics. If I'm
sending sensor readings, cheap but lossy may be just fine.
Etcetera. In practice, a relatively small number of patterns
is likely to emerge, with associated ontologies and semantics.

It's worth looking at how systems like JMS (Java Messaging
Service) coped with the problem of abstracting quality
of service (including reliability).