NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/unix/question/15044

Text File | 1992-12-30 | 9.5 KB | 193 lines

Newsgroups: comp.unix.questions
Path: sparky!uunet!spool.mu.edu!uwm.edu!ux1.cso.uiuc.edu!news.cso.uiuc.edu!uxa.cso.uiuc.edu!jdg41088
From: [e-mail] (Pickaxe)
Subject: Re: How many people are on the net?
References: <[e-mail]> <[e-mail]>
Message-ID: <[e-mail]>
Sender: [e-mail] (Net Noise owner)
Organization: University of Illinois at Urbana
Date: Wed, 30 Dec 1992 08:08:14 GMT
Lines: 181

[e-mail] (Steve Steinberg) writes:

>In <[e-mail]> [e-mail] (Gregory A Jerrett) writes:

>>I have been curious about this for some time.  Is there a rough estimate
>>out there anywhere of the approximate number of peopele/people who are
>>hooked up to the internet?

>A recent NY Times article said 3+ Megapeoples.

This may be much more of an answer than you really expect, or want for that
matter but here goes.

Reprinted from UIUCnet, the University of Illinois campus network newletter,
Nov 1991. vol 4, no 4.

******************************************************************

                  HOW BIG IS THE INTERNET?

How big is the Internet?  How many people use it and exactly how
is it used?  These are some of the common questions asked by new
and even experienced users as they ponder this vast and sometimes
overwhelming resource.  Unfortunately, there are no simple or even
direct answers.  How does one measure an internetwork consisting
of thousands of local, regional, mid-level and national networks
whose numbers are constantly growing and whose topologies are
constantly evolving--a "super-network", the management and
administration of which are so decentralized that some writers
have dubbed it "anarchy-net."  Although there is no single
authority to turn to for all the facts and figures about the
Internet, with the help of statistics gathered by  organizations
like the Network Information Center of the Data Defense Network
(DDN NIC) and Merit Incorporated (Operations Center for the
National Science Foundation Network [NSFNET]), we can at least
begin to take measure of this elusive network of networks.


How Many Networks?
------------------
One way to measure the Internet is to count the number of separate
networks it comprises.  Every TCP/IP network on the Internet has a
unique network address that must be obtained from a central
addressing authority, the DDN Network Information Center (for more
information on the DDN NIC, see the article and sidebar on page
3).  As of September 1991, the NIC had allocated nearly 37,000
network IP addresses.  Not all of these networks, however, are
connected to the Internet.   All organizations with TCP/IP
networks are encouraged to get their network numbers from the NIC
regardless of whether they plan to be connected to the Internet or
not. The greater part of this number actually represents private
TCP/IP networks without Internet connectivity.  When asked how
many of the 37,000 were actually part of the Internet, a NIC
representative reluctantly came up with the number 10,660 and
called the figure a "moving target" at best,  because new networks
are constantly being added to the register, and, more
problematically, occasionally large blocks of network addresses
are allocated to organizations that in turn reallocate them.  The
status of these blocks of addresses is unknown to the NIC.

We also know the number of networks expressly configured to access
the NSFNET national high speed T1 and T3 backbones. According to
Merit Inc., which has been maintaining statistics in this area for
many years, 516 networks announced their routes on the T1 in May
of 1989.  By May of '91 the figure had increased to 2,763, and in
October of this year the count was 3,556 on the T1 and another 869
on the faster, but less stable T3 backbone.


How Many Hosts?
---------------
While it is possible to specify with relative accuracy the number
of networks with Internet connectivity (or at least NSFNET
connectivity), the question of how many hosts (individual
computers with unique IP addresses) make up those networks is more
difficult to answer.  The absence of concrete data in this area is
due to the fact that the registration of new hosts on a TCP/IP
network is largely a matter of local concern.  For example, when a
new computer is attached to UIUCnet, the host name and IP address
of the computer is registered with the host manager at CSO.  The
host manager adds the information to several tables, and the
process more or less ends there.  There is no national database
containing the name and IP address of every host on the Internet
and thus no effective method of counting every host on every
network.  We do know, however, the minimum number of hosts on the
Internet as of October of this year.  An Internet Domain Survey
conducted by SRI International (former home of the DDN NIC)
reported no fewer than 617,000 individual IP hosts (as compared
with approximately 30,000 in April of 1988).  If this number seems
large, even more astonishing is the rate at which the number of
hosts has been growing in recent months. According to the same
survey, the number of hosts increased by 100,000 over the three
month period of July through September 1991 (or roughly 1000 hosts
per day).


How Many Users?
---------------
Some Internet hosts are single user machines like PCs, Macs and
engineering workstations.  Others are multi-user minis or
mainframes that can provide Internet access to tens, hundreds, or
even thousands of users (the CSO Unix mainframe uxa, for instance,
offers access to over 11,000 University of Illinois students).  At
the other extreme are hosts that do not provide any direct end-
user access such as those dedicated to the tasks of routing
traffic, gathering statistics, or performing other basic network
functions.  Thus behind each host there could be anywhere from
none to thousands of users.  Compounding  the problem is the fact
that there are many commercial, private, and restricted networks
that offer  limited  access to the Internet  (or  to regional
networks attached to it).   Should individuals on these networks
also be counted among the ranks of bona-fide Internet users?  In
short, nobody knows for certain how many users can access the
Internet, but whether the users with limited access are included
or not, it's probably safe to say that the number is, at the very
least, many times greater than the number of connected hosts.


How Many Packets?
-----------------
As the number of Internet networks, hosts, and users grows, so too
does the network traffic they generate.  For the last four years,
Merit Inc. has been counting both "packets"  (the basic unit of
information sent across a TCP/IP network) and bytes as they are
passed onto and off of the NSFNET backbones.  During the month of
September 1988, 202 million packets were passed onto the T1
backbone.  By September of '89, the number had increased to 1.5
billion packets per month and a year later the number had leapt to
4.5 billion per month.  September '91 packet counts for both the
T1 and T3 backbones were around 8.5 billion and 2 billion
respectively, nearly  fifty times the traffic of four years ago.

In September, Merit also began keeping track of the origins of
packets traversing the T1.  Merit's breakdown by country during
the month of September shows that 87% of the backbone traffic was
generated by 1758 different networks in the United States.
Another 3.3 percent of the traffic originated in Canada.  In
total, 2,958 networks from at least 33 countries sent packets
across the T1 backbone during this one month period.


How Is It Used?
---------------
Of the data gathered by Merit, perhaps most interesting are the
figures on how the network is used.  In May of 1991, Merit
released information on the most frequently used NSFNET
applications based on packet and byte counts.  The results are
shown in  figure 1.  Both in terms of packets and bytes, file
exchange (ftp) predominates, with electronic mail running a close
second. Interactive applications (remote login with telnet)
account for 17% of the packet count but only 6% of the byte count.
This apparent discrepancy can be explained if we understand that
TCP/IP packets can hold variable lengths of data.  During a telnet
session, characters are sent to and from the remote host one at a
time, so that the data area of each packet typically contains just
one byte.   Thus, the packet overhead of telnet sessions is high
while the number of bytes transmitted is comparatively low.
Applications like ftp (the TCP/IP file transfer utility), on the
other hand, pack as much data into each packet as possible,
requiring fewer packets with higher byte counts to accomplish
their task.


Still Growing After All These Years
-----------------------------------
How big is the Internet? As the foregoing has shown, by any
measure, it is rapidly increasing in size.   Its staggering growth
rate is such that even if one could obtain precise statistics on
the  number of users, hosts, and networks connected to it at any
given moment, they would undoubtedly be outdated before they could
be published.

-Lynn Ward

*********************************************************************

--
| Joe Gross          | Department of Electrical and Computer Engineering |
| [e-mail]   | University of Illinois at Urbana-Champaign        |
| [phone]     | "I've given up trying to stay ahead of the times, |
| NeXTmail welcome   | it's now a matter of how far I get left behind."  |