Re: SMTP Use Cases
2004-10-23 08:36:40
At 04:26 23/10/2004, Markus Hofmann wrote:
Hilarie Orman wrote:
There's an odd recursive aspect to applying OPES to SMTP, and I think
it is important to clarify why SMTP will benefit from OPES,
As I mentioned before - with OPES you can send parts of an applications
message to the callout server, you're not forced to send the entire
message. With SMTP, you would always have to sens the entire message. LEss
efficient. Do I miss anything here?
Dear Markus,
I am sorry but "I" miss something here.
- what has SMTP to do with this. SMTP has no callout server so there is no
way it can send parts or the whole message to a callout server.
- I suppose that what you mean is the that callout server may work on parts
of the content while most of the "value added" MTA systems would work on
the complete content. I do not see the added interest. Please remember that
SMTP permit many MTA to be involved on an traffic load balancing approach.
Far more efficient to work on a local content than to send and receive it.
Your proposed OPES is a sendmail, postfix, qmail etc. front-end adding to
their management complexity. Frankly I prefer developping a postfix module
or even to try to understand the qmail license and patch system or to see
what can be done with sendmail than to add management complexity ... All
the more than spam (which is an architectural problem) will only be solved
by an architectural solution, and since spam is a by-product of
TCP/IP¨+SMTP complexity, one will most probably find its solution in
simplification (please reread RFC 2775 on architecture) simplicity is of
the essence.
I dont say that you are wrong with your examples, services, OPES interest,
but that there is confusion and that you confuse protocol and content
massaging, synchronous and asynchronous processes, Host and users etc. and
that in some way you are right. But we need to know in which context we are
: in RTF 2775 context or in a RFD 2775 new permitted architectural context
- and then which one. To be modelized and documented.
jfc
Plus the additional - deployment/implementation specific - benefits Martin
and Alex mentioned.
-Markus
|
|