In message <19990519154215445.AAA324.273@jcochrane_ws>, Jeff Cochrane writes:
> > I use a Nokia 6110 with data connectivity running at 9600bps into my ISP
> > with great success. Compression can push it a bit, but 9600 works fine.
> We're trying to get a handle on costs.
> There's a small research center outside Kampala, but within the
> cellular area.
> One option they are considering is perhaps 5 cell phones each
> connected to 5 computers.
No, you don't wanna do that. Under NO circumstances!!!!
That's five times the basic rate! And everyone will uses it more or
What you want to do, is to wire this up as a LAN, and have a Linux box
act as the mail server/gateway using Taylor UUCP `i' protocol with
BSMTP batch gzipping (that way you get less overhead, compression,
what is most important, `i' protocol will send AND receive
concurrently, halving connection time).
That, however will mean you will have to make the ISP on the other
side offer Taylor UUCP dialin facilities with unbatching, ungzipping
and the `i' protocol. This is totally trivial (we have done it here
for years) but as most ISPs are dumbing down you'd have to do some
Then you run this under control, ie by hand, so you watch what is
happening so no hangups block the cell phone ($$$$$$).
> They think they can get 38,400bps data flow rates, but as some have
> indicated in this forum, that's likely unrealistic, a common
> misconception based on the deceptive way some modems present
> information to the user.
That's the country where the HF service said it would be cheap and
fast, right :-)-O?
> To compare the cost of cellular to the cost of a wireless leased
> line, we'd like to have a rough idea of what it costs to do email by
> cell phone. And there's also a quality difference to be considered,
> but for the moment we're just focusing on email.
I am still not satisfied as to why they can charge per KB on
wireless. EBONet in Angola doesn't.
> I assume you use your phone for more than just email, so we can't
> just look at your monthly bill. Instead we'd have to look at data
> flow rates, time online, etc.
Every cell phone has counters. Reset them, make a call and see how
long it takes.
> You say 9600. Any idea if that's your "connect" speed or if it's the
> actual flow rate of data?
I am quite sure that this is the nominal speed.
This is from Dieter Neuvians' message:
> Connection is with 9,600 bps, handshaking takes about 20 seconds,
> 100 kB mail (one message) uploading 3 minutes, 10 seconds,
> downloading 2 minutes 20 seconds.
Which comes to 4266 and 3413 Bps respectively. So let's assume an
average of 3900 Bps. But that was one 100k message. The smaller the
messages the worse it gets.
AND, mail software via TCP (Windoze in particular) will first do
POP/IMAP receiving (uncompressed and individual messages) and then do
SMTP (uncompressed and individual messages) or the other way round,
makes no difference.
Horribly inefficient when compared to UUCP/BSMTP, but then we have
discussed the virtues of UUCP some years in the past (ORSTOM using
UUCP via X25 uncompressed unbatched).
I am quite convinced that with a maximally tuned UUCP/BSMTP you can
double to triple the trhoughput.