Cell Relay Archive

[George Marshall <george@marshalls.org>]

Juhana R=E4s=E4nen wrote:
> =

>    Greetings,
> =

> I'd appreciate some opinions about UNI 4.0, whether or when it might
> be widely deployed, how compatible it is with UNI 3.1, is it in use
> anywhere, etc.
> =

> UNI 4.0 spec is now about one year old, but you don't (well, at least
> I don't) see much enthusiasm going on in the industry. Manufacturers
> don't bother to mention 4.0 in their product data sheets, but some
> protocol software houses do have 4.0 available. (But is anyone really
> buying it?) It seems to me that few really bother to take UNI 4.0
> seriously, as the current systems are more or less up and running 3.1
> without trouble. I might be wrong here and it's just going to take more=

> time, so opinions are welcome.

The lag time from completed spec to real presence in the market
(multiple, interoperating, fully released products) was a year+ for UNI
3.0, and even longer for 3.1 (of course, 3.1 added no new features, so
there was little reason to move to it).  UNI 4.0 does have a few new
features, but I'd expect a rather leisurely conversion in the industry
(i.e., not real soon) because few applications really depend on the new
features, and most vendors probably have other product feature
enhancements that matter more to their customers.
 =

> The reason I'm pondering this is that we have a research project going
> on involving UNI signalling. As we already have a prototype of Q.2931
> it would be natural to go on with 4.0 rather than 3.1, but the problem
> is that it would be nice to demonstrate interoperability with commercia=
l
> systems. UNI 4.0 spec doesn't state clearly how well it would work with=

> 3.1, so if anybody out there has tried out e.g. public UNI between a 4.=
0
> and 3.1 switch or 3.1 user signalling with a 4.0 switch, I would be ver=
y
> interested in your experiences.

Unless your research depends directly on some of the new 4.0 features,
stick with 3.1 for now; you will get interoperability with the maximum
number of other implementations that way.

George Marshall
--
george@marshalls.org