Re: [end points comm] OPES System

Hi,

rather than responding to individual emails, let me try to summarize 
the thread (while catching-up with queued emails...).
Goal: Tracing must allow the endpoints to detect that some OPES 
services were involved, and which party to contact in order to resolve 
possible problems. This has resulted in different views:
(a) One view is that this requires each individual OPES processor to 
support tracing. Policies can be used to authorize who can turn on 
tracing and who can see the results.
(b) Another view is that this requires only one trace per "OPES 
system", with an "OPES system" being the collection of entities that 
are trusted by the respective endpoint. This assumes that certain 
business arrangements/rules are in place, which will ensure that 
problems can be traced across carrier domains.
(c) One view might be that this requires only tracing per "OPES 
domain", with an "OPES domain" being the collection of entities 
operated by a single carrier.
Both, (b) and (c), assume that the carriers somehow have enough 
(internal) information to answer possible inquiries, e.g. they 
themselves must figure out which exact OPES processors have been 
involved when a specific user inquiry comes in.

For each of the views, it would be helpful to detail and to illustrate 
the entire tracing process showing an example scenario, including how 
a user would inquire about suspected problems and how this inquiry 
would be resolved. [It might also be helpful to learn from existing 
solutions to similar problems in other protocol areas, e.g. email etc.]
Since (b) seems to be the most "relaxed" view, let's start with this 
approach, and then see whether/how it solves real-life problems, 
whether the assumptions are valid, how it addresses the IAB 
considerations, and how differences to IAB consideration will be 
explained.
-Markus