[net.dcom] Cost of X.25 networks vs. cost of 1200 bps

covert@castor.DEC (John Covert) (03/25/84)

At 1200 bps, I can transmit (with an error correcting protocol)
approximately 5120 characters per minute, or per 47 cents (the
maximum cost of an interstate telephone call within the continental
U.S.).
 
This is 9.18 cents per kilocharacter.  Unless I'm mistaken, most
X.25 services charge much more than 9.18 cents per kilocharacter.
 
The basic line charge for an X.25 connection is drastically higher
than for a telephone; X.25 rates include both a time and a kilocharacter
component.
 
And note that the throughput I get with my error correcting protocol
is only 683 bps; a better protocol implementation should be able to
get that 56% line utilization up to over 80%.
 
And with a 2400 bps modem my kilocharacters should cost 4.6 cents.
 
Those 9600 bps access lines look real neat, until you find out that
the throughput of most X.25 networks is MUCH less.  I've measured
throughputs as low as 600 bps (gee, right there with my 1200 bps
modem and its error correcting protocol).
 
I'm not sold on the current X.25 offerings.  The potential is there,
but the economics don't justify it yet.
 
John Covert		...{decvax|ucbvax|allegra}!decwrl!rhea!castor!covert

dave@utcsrgv.UUCP (Dave Sherman) (03/26/84)

~|   From: covert@castor.DEC (John Covert)
~|   At 1200 bps, I can transmit (with an error correcting protocol)
~|   approximately 5120 characters per minute, or per 47 cents (the
~|   maximum cost of an interstate telephone call within the continental
~|   U.S.).
~|    
~|   This is 9.18 cents per kilocharacter.  Unless I'm mistaken, most
~|   X.25 services charge much more than 9.18 cents per kilocharacter.
~|    

Datapac (the Canadian network) costs anywhere from 21 cents to $2-3 per
kilopacket, depending on the two locations. A typical connection might be
50 cents. That's for a potential of 256,000 characters, or 0.19 cents
per kilocharacter, MUCH less than your 9.18 cents.

Additional Datapac charges depend on the type of connection you have. If
you dialup to the PDN, you pay an additional 4.5 cents/minute plus 35 cents
per kilopacket.

If you have a hardwired link to Datapac and your own PAD or X.25 support,
the connection costs a flat $120 (1200 baud) to $380 (9600 baud) per month.
And I have used Datapac (for law case searches) at 9600 baud; you do indeed
get throughput pretty close to that.

Of course, with current uucp handshaking you don't get anywhere near 256
characters per packet; but this could presumably be optimized (e.g., send
256 chars, get them *all* back to check, would cost the same as sending a
few chars and getting a checksum back).

I believe Telenet charges in the U.S. are roughly comparable to Datapac.
All prices above are in Canadian dollars ($1 = ~78 cents U.S.).



Dave Sherman
The Law Society of Upper Canada
Toronto
-- 
 {allegra,cornell,decvax,ihnp4,linus,utzoo}!utcsrgv!dave

piet@mcvax.UUCP (Piet Beertema) (03/28/84)

Your figures are correct if you take into account interstate connections.
However things get very different on international links:

On our transatlantic links we've found about the same effective speed
for 1200 bps phone links, viz. ~850 bps. The net cost here would be about
$0,3 per Kchar; however the transatlantic phone links are full of problems
causing the actual cost to be considerably higher.

On our experimental X.25 links within Europe I found an effective speed
of 2400 bps possible. Assuming the same speed could be obtained on
transatlantic links, the net cost would be $0.1 per Kchar. Besides it
is expected the X.25 links will give far less problems.

About the same story goes for phone vs. X.25 international links in Europe.
But within a country X.25 IS much more expensive. The morale: use X.25
only for international/intercontinental links.
-- 
	Piet Beertema, CWI, Amsterdam
	...{decvax,philabs}!mcvax!piet