[net.news] Just what DOES it cost?

preece@ccvaxa.UUCP (11/05/85)

Much has been made lately of the amount of net traffic and
the cost to backbone sites of maintaining the flow.  Some
very large figures have been reported (always as "I've heard
that X's phone bill is ...").  Anybody willing to volunteer
facts about the actual costs?

I had always guessed that much of the backbone traffic went
over long haul links established and maintained for other
purposes and therefore had no marginal costs.  Certainly there
are plenty of Usenet sites that have access to leased lines,
satellite links, microwave links, and whatever that have
some excess capacity that net traffic could piggyback on.

Does anyone know what the balance of trasmission methods is?
How much longhaul traffic goes by telephone and how much by
dedicated line?  What does a backbone site actually pay, per
month, for longhaul links?

Anybody with hard data and a desire for anonymity may send
data to me for summarization.

-- 
scott preece
gould/csd - urbana
ihnp4!uiucdcs!ccvaxa!preece

rees@apollo.uucp (Jim Rees) (11/12/85)

This info is two years out of date.  It's from when I was the sa at uw-beaver,
which is one of the backbone sites.

We didn't pay zilch for news.  We insisted that all of our feeds call us,
except for microsoft, which is a local phone call.  But this isn't to say
usenet was free.  We payed about $100 per megabyte for disk drives in those
days, and the news took up about 12 Mbytes, so that's $1200 amortized over
the life of the drive.  Chickenfeed.

The big cost was forwarding mail.  Our phone bill for mail was about $300
a month, and I would guess that over half of that was usenet replies.  The
frustrating part was that most of those replies were taking much longer
(about 3x) paths than they needed to take, because the originating sites
were too lazy to do any path optimization.  People used to complain about
the optimizations I did on their mail passing through my site.  To them
I said, "Life's a bitch."

You can at least double those numbers now, because volume on the net has
at least doubled.  Interestingly enough, the doubling period has remained
fairly constant at about every two years for the five years I've been on
the net.  Has anyone else been keeping track?

I'd love to see decvax's phone bill.  So would Armando's boss, but I won't
tell, and he's fled to California now anyway.

richl@lumiere.UUCP (Rick Lindsley) (11/14/85)

In article <1300024@ccvaxa> preece@ccvaxa.UUCP writes:

    >Much has been made lately of the amount of net traffic and
    >the cost to backbone sites of maintaining the flow.  Some
    >very large figures have been reported (always as "I've heard
    >that X's phone bill is ...").  Anybody willing to volunteer
    >facts about the actual costs?

Ok. I'm the postmaster at Tektronix and I have been asking that very
same question. Especially since there are people in this group who seem
to doubt that news backbones really pay all that much. Since I and
our news administrator here are constantly finding we need to justify
these costs internally, I have the information right here.

I just asked our telecommunications folks what our phone bills
have been for the last three months. Tektronix (West Coast) talks
long distance to decvax (East Coast) and two other West Coast sites
to exchange news and mail. We talk to a number of other sites around
the country to exchange mail only. We speak to a number of local (Portland)
and Tek-local sites for both news and mail (so no additional cost there,
unless you want to count in the cost of the numerous local area networks
here at Tek, which would probably exist anyway).

This is not an "I've heard". This is not an "I guess". This is the total
cost for two dialers for news, mail, and CSNET. Over the last 3 months,
we have had a "low" of $5300, and a high of $7000 (rounded off to the
nearest hundred dollars). We talk long distance only after 5pm and before
8am. During the day, we make only local calls (though *other* sites are
free to call *us* -- but of course they seldom do for the same reason).
And the cost is on the increase, despite batching, compression and just
about anything else you can try.

Now you are of course saying, aha, but that is not just news. No of course
not. Only 80% of that is news. By cost, about 13% of that is CSNET (also
a call to the East Coast) and only 7% is mail-only sites. The exact figures
vary, of course, from month to month, but news is consistently the hog on
the dialers.

    >I had always guessed that much of the backbone traffic went
    >over long haul links established and maintained for other
    >purposes and therefore had no marginal costs.

Hahahahahahahahaha. Nope. Perhaps we are backward, but I think we are
more likely "typical". We have one dedicated line to a Tek machine in
central Oregon. We have extensive networks within Tek (which through
the use of microwave dishes actually expand to reach about 30 miles, at
their longest point).  Everything external is phones. We are installing
an X.25 link -- but that is not peanuts either (and doesn't work yet
anyway).

    >Does anyone know what the balance of trasmission methods is?
    >How much longhaul traffic goes by telephone and how much by
    >dedicated line?  What does a backbone site actually pay, per
    >month, for longhaul links?

Were it not for the local nets, Tektronix would go broke just
delivering news to its internal sites.

You can see what the cost is to a site like Tek when groups like
net.flame or net.bizarre get created. Ideally, a backbone should not
have to restrict any newsgroups to the sites it passes to. But with
monthly phone bills in the $6000 area, it gets harder and harder to
justify those costs. I applaud the efforts to clean up the net.

You are thinking, "well, why don't they install some cheaper
alternatives?" Well here's why. Because any alternative will require
planning and justification. And while maybe people will overlook a
$70,000 yearly expenditure in increments of $6000, they WON'T ignore,
say, a $50,000 one time expenditure. While they are looking at this
$50,000 expenditure, they eventually ask the question, "what if we just
eliminate news entirely?" And then they realize they can achieve a
$50,000 yearly SAVINGS, as the phone bills drop to maybe $1500/month...
Most managers are interested in short-term gains or losses that affect
them for the immediate budget (sad but true).

So if any of you complainers out there want to be a backbone, that's
great!  Plunk down your $6000/month and join the ranks. Maybe Tektronix
can step down and be a parasite off of somebody. I'll certainly lose a
lot of headaches at budget time.  Alternatively, send a contribution to
the backbone of your choice. I'm sure they'll appreciate it.

Hoping this opens some eyes (but not holding my breath),

Rick Lindsley
Postmaster@tektronix
richl@tektronix.csnet
...{decvax,ihnp4,ucbvax,allegra}!tektronix!lumiere!richl

dmcanzi@watdcsu.UUCP (David Canzi) (11/17/85)

Unlike some people, I don't believe that I have an inherent God-given
Natural Right to run off at the keyboard at other people's expense.  I
would favour a system where the poster pays to post an article, but I
think there are technical and legal problems with implementing such a
system.

To satisfy my curiosity, can anybody provide a reasonable estimate of
how much it costs other people when I post a 100-line article?
-- 
David Canzi		"For every bloke who makes his mark, there's half
			a dozen waiting to rub it out." -- Andy Capp

lauren@vortex.UUCP (Lauren Weinstein) (11/20/85)

The "poster-pays" concept has the fundamental philosophical
problem (leaving aside the very significant technical problems for now)
that money does not necessarily equal valuable messages.  A site with
lots of bucks to post messages doesn't necessarily have people with
anything useful to say.  A site with no money for postings
may have people who could post many useful items, answer queries
in an intelligent manner, etc.  This is part of the reason that
editors exist.  A magazine like "Time" doesn't just charge a fee
and say "we'll print any (legal) article if you pay the price."  Instead,
they judge articles on their merit and try to produce a useful
publication for the benefit of their readership.  As the readership
changes, complains about, or praises certain materials, the content
of the magazine gradually changes as well.  Readers who don't like the
content of Time can also go read other magazines with other editors
and other content priorities.

Saying that you can't post unless you can pay would be a big mistake.

--Lauren--

dmcanzi@watdcsu.UUCP (David Canzi) (12/12/85)

In article <856@vortex.UUCP> lauren@vortex.UUCP (Lauren Weinstein) writes:
>The "poster-pays" concept has the fundamental philosophical
>problem (leaving aside the very significant technical problems for now)
>that money does not necessarily equal valuable messages.  A site with
>lots of bucks to post messages doesn't necessarily have people with
>anything useful to say.  A site with no money for postings
>may have people who could post many useful items, answer queries
>in an intelligent manner, etc.  

My estimate of 10 cents per line (probably pessimistic) would exclude
few, if any, people from posting, but it would make them think twice.
And if some rich fool posts reams of garbage, so what?  He's paying the
cost.  And news-reading software can automatically skip his articles,
so nobody need even be bored by him.  Answering queries would most
likely be done via mail (free or cheap), rather than posting.

As I understand it, charging people for posting and reimbursing
backbone sites can't be done efficiently without an organization to
collect and distribute the money, and that organization would be a
potential target for lawsuits for things that get posted over the
network.  That's the problem with "poster pays".

>                                This is part of the reason that
>editors exist.  A magazine like "Time" doesn't just charge a fee
>and say "we'll print any (legal) article if you pay the price."  Instead,
>they judge articles on their merit and try to produce a useful
>publication for the benefit of their readership.  As the readership
>changes, complains about, or praises certain materials, the content
>of the magazine gradually changes as well.  Readers who don't like the
>content of Time can also go read other magazines with other editors
>and other content priorities.

Usenet isn't a magazine, though, and I can't think of any reason to try
to make it like one.  Usenet is not like any other mass medium: its
content is provided entirely by its readers and not controlled by any
government or corporation.  It's a mass medium which is not restricted
to carrying "safe" opinions and "official truths".

The problem with Usenet is the unreimbursed cost to the backbone sites,
which can be reduced by limiting the amount people can post or charging
the users and reimbursing the backbone sites.  Moderators might be
needed to approve truly worthwhile but large postings, such as
sources.  But, since a moderator can be sued for things he allows to be
posted, giving him complete control over what can or can't be posted
would be a mistake.
-- 
David Canzi

Law of the Yukon:  Only the lead dog gets a change of scenery.