return to first page linux journal archive
keywordscontents

Penguin's Progress

Where Have the Nets Come From?

by Peter H. Salus

On September 2, 1969, the first piece of the ARPANET was installed in Len Kleinrock's lab at UCLA. A month later, the second went into Doug Engelbart's at SRI in Menlo Park, CA.

As of the end of last year, there were over 68 million hosts attached to the Internet, and those hosts average four users each. A quarter of a billion folks all over the world.

I really do mean all over.

The latest analysis by MIDS (http://www.mids.org/) of the data assembled by Network Wizards shows host sites in over 200 countries or territories--far more than the members of the UN.

MIDS noted a year ago that while the Internet was growing, the growth rate, which had shown a factor of two or better (i.e., the size of the Internet doubled each year) for just over a decade, was beginning to slow. The four most recent data points (January 1998, July 1998, January 1999, July 1999) show the growth rate to be 1.5 now. This is still a lot.

The 68 million of January 2000 is projected to become 100 million this year, 150 million at the end of 2001, and 225 million at the end of 2002.

If we assume the average will stay at four, that's a billion people in less than three years.

Why Has the Growth Been So Incredible?

I think there are several factors: first, the advent of the Hayes modem. The progress from 1969 to 1975 brought us from the half-rack that was used in the original ARPANET to the shoebox of the acoustic modem. The subsequent reduction in size (and cost) was vital. Second, there was the ``miracle'' of the desktop computer--the Apple, the LISA and the IBM PC (then the XT and the AT). I'm not trying to be detailed or expansive here: I know about the Osborne and the Kaypro, etc.

With these two developments, you could have a machine at home, plug it into your phone line and communicate with others. That is, if your line had modular plugs; otherwise, you either bought a plug with four circular pins on one end and an RJ-11 on the other, or you wired into your line at the wall and attached a cable with an RJ-11 at the end--both were ``illegal'' acts where Ma Bell was concerned.

In October 1981, there were 213 hosts on the Net. In February 1986, there were 2308. From 1986 to 1997, the number of hosts doubled every year (the actual slope of the line is 2.176). But since January 1998, as I noted above, that growth rate has fallen.

I wouldn't worry about this.

Every technology goes from a lab to being the toy of an in-group to being a utility. When it first becomes a utility, people rush to buy and participate, then there is a slackening.

But don't fret. In July 1999, Malawi, Kiribata, Eritrea and Somalia joined the Internet. In January 1999, they were preceded by Sao Tome, Samoa and Tuvalo. There are so few additions because there are so few countries left unconnected.

The scale of that penetration and interconnection is quite important. MIDS' analysis of the July 1999 data lists over 250 geographic entities. I don't use ``countries'', because Guam, Antarctica and the Vatican City (for example) aren't countries. But that's still more places on the Internet than countries belonging to the United Nations.

Doc Searls pointed out some months ago that about a third of the Net servers were powered by Linux. As the Internet grows, Linux's use will grow with it. There's no way that Windows (which comes in third, after UNIX and Linux) will overtake Linux in the Internet back rooms--nor, I imagine, in the corporate intranets.

I'd put my money on technology by Torvalds, Gosling, Ousterhout, Allman, Wall, et al., from companies like Transmeta, VALS, Red Hat, Caldera; not on Wintel.

Peter H. Salus, the author of A Quarter Century of UNIX and Casting the Net, is an LJ contributing editor. He can be reached at peter@ssc.com.