Mark Mentovai <[EMAIL PROTECTED]> wrote:
>
> I use "should" in the same manner that it is used in the documents which
> define the very standards and practices over which we are arguing. In order
> to be a good 'net neighbor, an MTA (note that I am not singling any MTA out
> here) should not open 25 SMTP connections to the same host to transfer the
> same message specifying a different destination address each time when it
> can just as easily open a single connection and specify 25 destination
> addresses.
[...]
> Is it as fast as possible? In the situation above, what I suggest should
> happen is actually faster and makes better use of network resources than
> qmail's current implementation.
"Measure; don't speculate" -- DJB.
Real world tests show that in most instances, qmail will deliver its
multiple copies of an email with multiple recipients at the same MX
faster than sendmail will with its single-copy, multiple RCPT strategy.
And qmail uses less system resources in the process. Also, by not tying
up an SMTP session for an extended period of time, one could argue that
qmail is more network-friendly for it.
The problem with re-using the same SMTP session for multiple messages, etc,
is the high-latency inherent in the protocol. DJB found an easy way around
that.
Charles
--
-----------------------------------------------------------------------
Charles Cazabon <[EMAIL PROTECTED]>
GPL'ed software available at: http://www.qcc.sk.ca/~charlesc/software/
Any opinions expressed are just that -- my opinions.
-----------------------------------------------------------------------
