On Sun, 16 Sep 2001, Pavel Kankovsky wrote:

> > Depends on the client. qmail-remote does not use PIPELINING even if the
> > other side offers ist.
> 
> Indeed. But I want to see numbers. Profile, don't speculate! ;)

Still in the works, not a priority item here, SMTP PIPELINING is fast
enough for me.

> > > How is the difference affected by SMTP pipelining?
> > 
> > Approximately the same. SMTP PIPELINING and QMQP get closer to each
> > other the more recipients the mail has.
> 
> Approximately the same == not at all? If the answer to the previous
> question was not ``not at all'', it would make some sense.

Depends on what you compare it against. Ask your question again
precisely please.

> > > How is the difference affected when multiple concurrent connections are
> > > used (amortizing RTT-induced delays in SMTP)?
> > 
> > Small, that's why qmail spawns multiple qmail-remote clients if it has
> > more than one mail to send.
> 
> Excuse me? Do you want the say ``the difference would be small'' or ``the
> difference would not be affected much''? (I suppose you meant the latter.)

Parallel connections can fill the idle gaps that round trip delays
cause, however, they require additional overhead for connection
establishment -- which you do not have for a single SMTP PIPELINING
connection (maildirsmtp).

> > Personally, my mails have up to 10 recipients, more only in exceptional
> > cases, uusually only 1 recipient.
> 
> I do not send many multi-recipient mail (esp. across a modem link) either.
> But it does not matter. You yourself showed (E)SMTP + PIPELINING can be as
> fast or even faster than serialized QMQP (yes, serialized: it is not a
> good idea to run too many paralel connections through a slow modem link,

It doesn't matter, but keep in mind TCP is not fair, so the first
connection you open may starve those subsequently started.

> is it?) when the number of messages is something like 4 or 5 or higher. On
> the other hand, the for a smaller number of messages, the differences
> would be too small, measured in seconds in the worst case. So, where is
> the *huge* benefit of QMQP?

Ask that the inventor.

-- 
Matthias Andree

"Those who give up essential liberties for temporary safety deserve
neither liberty nor safety." - Benjamin Franklin

Reply via email to