Re: discussion priorities

[Henry Lowe <hlowe@omg.org>]

Jim,

OK, I'll have to look at the Reliable Messaging (RM) spec to see if 
what you are talking about does the job.  

The reasons I'm concerned about this are
1. in the past, being able to send a message from one partner 
   to another using multi-hop with different transport protocols 
   on each hop, particularly for SME support, was seen as very 
   important, and 
2. if you don't have ebXML routing (and, as David Webber was 
   suggesting, left routing to the network), IMHO, you don't need 
   the type of RM Fujitsu is proposing as there are loads of 
   protocols out there that already have it (why re-invent it, 
   unless, of course, you need it for the end-to-end guarantee 
   over multi-hop routes managed by ebXML Messaging Service).

Best regards,
Henry
------------------------------------------------
At 09:40 PM 08/22/2000 -0700, Jim Hughes wrote:
>For simplicity in the RM discussions, we tabled discussions of multi-hop 
>networks.
>
>However, we did define a Routing Header, which defines the Messaging 
>Service Handlers used in a particular link pair in the traversal. If we 
>mandate the use of Routing Headers (even for trivial one-link-pair 
>traversals), then this could hold the information...
>
>Jim
>
>At 05:28 PM 8/22/00 -0400, Henry Lowe wrote:
>>All,
>>
>>I didn't see anything in the Messaging saying how routing is
>>handled.  One of the things we've talked about since the
>>beginning is multi-hop routing, e.g., customer at A orders
>>widget W from vendor at V bu the message goes via a portal P
>>which means it travers two links (one of which might be HTTP
>>and the other MQSeries or CORBA).
>>
>>I included this in my comments, but didn't propvide text.
>>You would probably use source routing for this sort of thing
>>where the route(s) would be contained in the TPA.  But some
>>how these have to get into the Header of the Message so A
>>knows to send the message to P and P knows to send it on to V.
>>Also need to have a place in the headers to record it's
>>actual wild and wooly route through the Internet for traceability.
>>
>>Or is this all there and I just missed it?
>>
>>Best regards,
>>Henry
>>-
>