Plain Text Please?

[Tony Wells]

Tim wrote:
>When adding this to the plain text it mean
>that between twice and three times the essential amount of data was
>being shifted.

I'm in danger of pushing this issue off topic. But I couldn't stop myself
from responding, Tim.

As a subscriber to a number of mailing list, I come across the debate on
formatted mail frequently. Such mail sometimes seems to evoke extraordinary
reactions. One frequently-quoted reason for the anti-formatting is the issue
of extra bandwith required to transfer mail.

I did some research a year ago which seemed to show that formatted mail was
not really an issue on network bandwidth.

When writing some smtp/pop3 client/server software I wanted to ensure the
mail was being pumped as fast as possible. So I did some measurements.

I noticed that a large number of mail servers had measurable inter-message
delays. Eg delays in responses to the initial mail header requests and a
delay in response after the client sent the mail terminator. The result of
this is that for messages below about 2k at 28kbps there was significant
latency . This latency was roughly equal to  the transfer time of the
message body itself.

There were further delays introduced by the mail client which also had to
fetch the message from its internal o/g message queue before sending and
similar delays on receiving. Clearly during these delay, other network
traffic can flow.

On circuits faster than 28kbps the latency of receiving mail servers and the
sending servers (including high capacity smtp server) generally exceeeded
that transfer time of the message. The latency is due to internal processing
& lookups on the mail server.

This bears out the general rule of thumb that on fast circuits, if mail
throughput needs to be speeded up,  it is not sufficient to increase network
bandwidth. The processing speed of the servers has to be increased.

So, my conclusion is that the impact of widespread formatted mail on mail
thruput is negligable. If however a high-capcity mail server improves its
message thruput with an upgrade, extra bandwidth  will be needed or other
traffic will be degraded. Again this is a known phenomena.

Regards,

Tony Wells




---------------------------------------------------------------------
The VNC mailing list - see http://www.uk.research.att.com/vnc/intouch.html
---------------------------------------------------------------------