Load Balancing with Postfix [and SpamAssassin]

Michael Torrie torriem at chem.byu.edu
Sat Jan 21 00:27:04 MST 2006


On Fri, 2006-01-20 at 16:58 -0700, Ross Werner wrote:
> I think either you or I is confused as to what "best effort" is.

I have been using the term in more of an ad-hoc fashion.  I understand
your protestations a little better.  I termed e-mail best effort,
meaning the mail server tries really hard to deliver the e-mail, but
there are no guaranties that it will be delivered successfully in a
timely manner.  If the mail is delivered successfully, we can be sure
that the message was transmitted in its entirety to the remote server
and you hear nothing back.  This is not a a guarantee that the recipient
received the e-mail though, merely that the other server claimed to
receive it.

And I referred to TCP/IP in a similar sense, in that a tcp connection
will continue to retry packets until an acknowledgment is received or
until a timeout occurs.  Or if the remote machine is not available or
not accepting packets, an icmp packet reports failure.  To me that is
best effort.  Whereas IP alone just simply transmits packets with no
regard to success or failure.  IP packets do, as you alluded to, contain
a checksum that can guarantee (relatively speaking) that if you have
received a packet it is correct and in its entirety.

It does make for interesting arguments when we haven't even agreed on
the terminology.  It's my fault of course, but then again not telling
you my definitions gives me the egotistical edge.  :)  It's the secret
to politics.



More information about the PLUG mailing list