Not sure what this proves ...

[isham-research.freeserve.co.uk at pop.pol.net.uk]

Delivery-date: Mon, 04 Dec 2000 05:00:42 +0000
                Sun, 3 Dec 2000 20:47:50 -0800

Delivery-date: Mon, 04 Dec 2000 04:56:11 +0000
                Sun, 3 Dec 2000 20:45:53 -0800

Delivery-date: Mon, 04 Dec 2000 04:52:05 +0000
                Sun, 3 Dec 2000 20:45:04 -0800

The Internet has no FIFO guarantee - traffic passes across it in
unpredictable ways and there are no Service Level Agreements binding
across it.  If you send two messages, they might arrive in any order
or indeed not at all.

What _is_ interesting, however, is that the above headers give us an
approximation of the list server's external performance.  Very
roughly, it seems to take Dan's server four minutes to deal with an
empty message.  "Empty" in this context is relative - even a null
message like these is still 2.1KB, largely as a result of mailman's
pointless X- headers.  That's about 3MB of bandwidth out of the
server per message sent, plus the content and any untrimmed rubbish.

So the untrimmed messages I was complaining about recently EACH wasted
approximately nine minutes of Dan's bandwidth.  It's obvious - at least
to me - that you only need a few of those per day to seriously impact
list bandwidth and response time.

I hope I can now be excused for having a 'bad attitude' to bandwidth
wastage.  Conscientious trimming will SERIOUSLY improve the list's
performance.

--
 Phil Payne
 Phone +44 7785 302803   Fax +44 7785 309674