RE: Synchronous Messaging Proposal - Some Thoughts

["Burdett, David" <david.burdett@commerceone.com>]

Michael wrote ...

> >    Proposal: OnceAndOnlyOnce messages should be required to use a
> > synchronous transport medium. The acknowledgement should arrive on the
> > same connection as was used to send the original message.

... and Chris said ...

>>>I think that this deserves some discussion. <<<

So here's my $0.02c worth ...

Basically I disagree with this restriction. Using a synchronous transport is
fine, but there is no need, IMO, to restrict reliable messaging to this type
of transport. The key elements of any reliable transport are the following:

1. Resending of an original message by the first party if no acknowledgment
is received within a specifieid time
2. The sending of a response message (whether an ack or a business response)
to the original message by the second party to demonstrate to the first
party that the original message was received
2. Resending of the response message by the second party whenever the same
message was received more than once, and
4. Duplicate filtering by the second party so that the ultimate recipient of
the message only receives it once.

This works equally well whether you are using SMTP or HTTP, except that the
Timeouts on SMTP would probably be a bit longer ;)

Otherwise I agree with Chris's comments - again ;)

Thoughts?

David
-----Original Message-----
From: christopher ferris [mailto:chris.ferris@east.sun.com]
Sent: Tuesday, December 12, 2000 4:42 AM
To: ebxml-transport@lists.ebxml.org
Subject: Re: Synchronous Messaging Proposal - Some Thoughts


I've moved the discussion to trp list where it belongs (IMHO).

Please see my comments below.

Cheers,

Chris

Prasad Yendluri wrote:
> 
<snip/>
> 
> Michael Joya wrote:
> 
<snip/>
> >
> > 1.  Make the clear distinction in the TRP Message specification between
> > acknowledgement messages and reply messages. Ack messages are sent by
> > the MSH layer to another MSH layer and are central to the ebxml
> > messaging protocol. Reply messages are sent on behalf of the application
> > or service and convey some meaning exterior to the ebxml messaging
protocol.

ABSOLUTELY! 110% agreement there.

> >
> > 2.  Acknowledgement messages deserve the most reliable delivery method
> > available. In a synchronous transport protocol such as HTTP/HTTPS, this
> > is readily available and is the immediate connection upon which the
> > request was made.

synchronous response is not necessarily the most reliable means of
getting an ack. The network connection can be severed before the
ack is delivered. 

> >    Proposal: OnceAndOnlyOnce messages should be required to use a
> > synchronous transport medium. The acknowledgement should arrive on the
> > same connection as was used to send the original message.

I think that this deserves some discussion. 

> >
> > 3. One-way protocols (SMTP) are inherently bad choices for building an
> > additional layer of reliability upon. However, because they often have
> > their own assured delivery mechanisms, they make good choices for
> > "BestEffort" messages.
> >    Proposal: BestEffort messages should be required to use an
> > asynchronous transport.

I don't necessarily agree with this. There's no reason why BestEffort
can't leverage a synchronous transfer layer (HTTP/S).

> >
> > Dick, would 2 and 3 infringe on DOE requirements?
> >
> > --
> > // mike.joya@xmlglobal.com
> > // XML Global
> > // POC Project Team - ebXML
> > // Vancouver, Canada
> > // 604.717.1100 x230