I'm looking for the following statistics on Internet backbone usage:

Total internet backbone capacity (in, say, gb/sec)
1990 	1991	1992	....	2001	2002E	2003E	2004E	2005E

Total internet backbone usage (same metric)
1990 	1991	1992	....	2001	2002E	2003E	2004E	2005E

Wholesale price per gb/sec
1990 	1991	1992	....	2001	2002E	2003E	2004E	2005E

Investment cost per incremental gb/sec
1990 	1991	1992	....	2001	2002E	2003E	2004E	2005E

---

Request for Question Clarification by vercingatorix-ga on 03 Dec 2002 12:40 PST

After some research I've discovered a major problem. Most of the data
you request is not collected by anybody. The companies that know are
not required to report, and thus do not. I've found a few rough
estimates from reputable academics, but most of those are merely
percentages, with very broad parameters, and differ greatly depending
on where I read them.

Thus, any number I generate, regardless of the source, could be hugely
incorrect.

Would you rather get a very loose, broad-based estimate, or more
specific data relating to a small sample of the market?

V

---

Clarification of Question by doctorwes-ga on 04 Dec 2002 06:22 PST

That is very interesting information in itself.

Under the circumstances, a loose range of estimates would be
acceptable; but a pointer to "more specific data relating to a small
sample of the market" might also be useful, depending on what it was.

---

Clarification of Question by doctorwes-ga on 04 Dec 2002 09:24 PST

It would also be useful to know who are "the companies that know".
Does this imply that the raw data required to compile these statistics
does exist, but is private; or is it the case that the raw data is not
being collected at all?

---

Request for Question Clarification by vercingatorix-ga on 04 Dec 2002 11:10 PST

ISPs and others who provide services know the usage. They are also the
ones likely to know the capacity. They are not likely to share this
data en masse with their competitors. I wouldn't be surprised if a
governmental entity or industry group collected the data, but
something like that is going to be aggressively guarded. Very private.

Prices differ greatly from region to region and depending on the
nature of the purchaser.

And as far as investment cost, I don't know if I can get that, even in
a very rough form.

Regardless of the route I use, because of the inaccuracies of the
estimate and the wide spread of prices, I'm afraid that any industry
average will have no real predictive value. Of course, if you just
want to know, the numbers may still be of some value.

Please let me know your intentions. If you want the best I can dig up
on these statistics, I'll give you that. If the data won't suit you,
leave the question around for another researcher. Perhaps someone else
has access to better data.

---

Clarification of Question by doctorwes-ga on 04 Dec 2002 14:13 PST

The best that you can dig up on the first three statistics would be an
acceptable answer, provided you give references and some (subjective?)
degree of uncertainty.

If you can explain why investment cost cannot be estimated, even
roughly, that would constitute an answer.

The fact that this question is so difficult to answer is, in itself, a
very useful thing to know. That is not what I expected at all.

---

Request for Question Clarification by vercingatorix-ga on 04 Dec 2002 15:08 PST

OK. I'll get to work on it. Be advised, this may take some time.

V

---

Clarification of Question by doctorwes-ga on 04 Dec 2002 16:28 PST

That is fine.

---

Clarification of Question by doctorwes-ga on 12 Dec 2002 12:26 PST

I will post a clarification shortly. Sorry for the delay.

---

Clarification of Question by doctorwes-ga on 16 Dec 2002 09:55 PST

Here is a question clarification as requested by enderwiggen-ga.
However, I will accept any answer that corresponds to a reasonable
interpretation of the question.

'backbone': t1 to oc3 lines

'usage': the sources suggested seem fine

'price': monthly price for a given type of connection is fine

'investment cost': by backbone providers

I think you or any researcher attempting to answer this, will find a
lot of information at http://www.caida.org. I'd have answered this
myself, but I have other commitments currently.

Good luck.

Just to let you know, several researchers have been working on your
question, it has not been forgotten or ignored.

Most of this information is available.  There are some questions /
problems with the request though.  For the purpose of answering your
question you have to define what you mean by 'backbone'.  The best way
in my opinion to do this would be

1.  Line quality (t1 to oc3 lines?)      (cheapest way to find out I
would say)
2.  Referents to the prime name servers  (investigate via Internic)
3.  Traffic on all class A networks as defined as networks using class
A IP addresses.

The definition applies to your questions in the following ways:

Total internet backbone capacity (in, say, gb/sec) 
As the saying goes a chain is only as strong as its weakest line. 
Therefore backbone capacity depends on where you truncate the
backbone.  On the high end you want to look to lucent's work with
optical router capacity as it is the highest single 'point' of data
transfer.
 
Total internet backbone usage (same metric) 
1990  1991 1992 .... 2001 2002E 2003E 2004E 2005E 

Same problem only more so.  As it may be functionally assumed that all
internet traffic crosses the backbone by any of the above definitions
you may refer to this as total internet usages.  Reliable statistics
on the including in some cases full color maps from Lucent, Sprint and
from a different approach 'media metrics.'  The third perhaps
overlooks non-us server traffic.

Wholesale price per gb/sec 
1990  1991 1992 .... 2001 2002E 2003E 2004E 2005E 

Not really a meaningful statistic.  In 2001 most commercial bandwith
was sold on a monthly basis not per second.  Data transfer in excess
of the total per server was typically charged at an escelated rate.
Consequently there really is no /per second price.  As for average
wholesale price of bandwith this can be obtained from find SVP, or any
one of several public advocacy groups.

Investment cost per incremental gb/sec 
1990  1991 1992 .... 2001 2002E 2003E 2004E 2005E

Investment by whom?  Backbone providers?  Class A lease holders? 
Total US investment in technology?  Investment by ISPs?  Personal
investment etc.


Hope this helps