Rich Pieri <rich.pieri@prescienttech.com> wrote:
>
>Dave Barr writes:
>
>>Rich Pieri wrote:
>>> Your logic is flawed, Dave. One instance of sendmail does the same job
>>> as 20 instances of qmail-smtp.
>
>> And you say Dave's logic is flawed?
>> How can you possibly support this statement?
>
>Assuming sender is sending 20 messages to recipient:
>
>Case 1: sender running sendmail.
The situation we were discussing was "sender running qmail, recipient
running sendmail vs. recipient running a modular MTA". If you want to
change the rules, please do so before stating a conclusion based on
the new rules.
>Sender generates one SMTP session for all
>20 messages. Recipient requires one instance of his MTA to handle the SMTP
>session, for a total of 1 instance.
>
>Case 2: sender running qmail. Sender generates 20 SMTP sessions, one per
>message. Recipient requires one instance of his MTA for each SMTP session,
>for a total of 20 instances.
Note: if recipient is not running a monolithic MTA, it will only need
20 instances of its "smtp server" function.
>Both cases deliver 20 messages from sender to recipient. Case 1 has one
>instance of sendmail; Case 2 has 20 instances of qmail.
There's no such thing as an "instance of qmail". In this case you're
talking about instances of qmail-remote.
>One instance of sendmail does the same job as 20 instances of qmail.
Yes, in this scenario that's true (ignoring the meaninglessness of
"instances of qmail"). It also takes the single sendmail about 20
times as long. Which is precisely *why* DJB designed qmail to do
parallel delivery.
-Dave
Follow-Ups:
References:
|
|
