Re[2]: greatings pampered elite -- the Column

[David Farber <farber () central cis upenn edu>]

Posted-Date: Wed, 08 Sep 93 22:31:54 PST
Date: Wed, 08 Sep 93 22:31:54 PST
From: "Bob Metcalfe" <Bob_Metcalfe () ccgate infoworld com>
To: farber () central cis upenn edu (David Farber)
Subject: Re[2]: greatings pampered elite






Dear Dave,


Here below is an EARLY DRAFT of my latest Internet column.  The actual column 
was edited some and is more accurate and better written.  For your list with 
that disclaimer.  The actual column is burried in our production system 
somewhere out of my reach right now.


Stay tuned, and thanks for asking,


/Bob Metcalfe


-------------------


DRAFT TWO


InfoWorld / From the Ether / Bob Metcalfe


Modernizing the Internet for commercial use


Perhaps you've stayed away from the TCP/IP Internet because it's mostly for 
academic Unix.  Perhaps TCP/IP sounds to you like an admission of damning drug 
test results.  But now the Internet reportedly connects 20 million computers 
and is doubling every year.  So your DOS and Macintosh PCs are probably already 
on the Internet, and if not, then soon.
    The Internet has since 1969 been a research project, funded through DoD's 
Advanced Research Projects Agency (Arpa) and the National Science Foundation.  
The Internet's first generation was called Arpanet, and it connected mainframes 
and minicomputers cross-country at 50 kilobits per second.
    In 1983, a new generation of network protocols, TCP/IP, were installed on 
the Arpanet mainly to interconnect proliferating megabit-per-second LANs of 
Unix workstations.  In this second generation, Arpanet became Internet and 
since then has proven a very good model for how workstations and PCs (if 
there's a difference anymore) should be networked.
    The newly formed Internet Society projects, with tongue in cheek I assume, 
that the Internet will connect every human on earth by 2001.  Well before then, 
however, the Internet will outgrow TCP/IP.  So the Society is working toward a 
major protocol upgrade -- a third generation? -- so the Internet might become 
what the Clinton administration calls National Information Infrastructure (NII).
    But wait a minute.  Since cutting over to this third generation Internet is 
going to be a big deal, let's take more time to get it right.  And I don't mean 
to get the Internet right to be "infrastructure" -- a code word that many 
Internauts (but few Clintonistas) take to mean "funded by the government."  I 
mean get the Internet right for commercial use, so it can help make the world 
go around.
    Networks joining the Internet need blocks of addresses for their attached 
computers.  These addresses are free and sell like hotcakes, so the Internet is 
running out of addresses.  Several schemes for enlarging the addresses used in 
the Internet are being considered by the Society.  By the time a scheme is 
chosen and a migration plan worked out, software in tens of millions of 
computers and the switching systems between them will need upgrading in one of 
the biggest cutovers since Great Britain decided to drive on the right side of 
the road.
    My point is that addresses are not all that needs upgrading if the Internet 
is to be developed for commercial use.  I think the next generation should be 
capable of, among other things, fully exploiting ATM, serving individual users 
via ISDN, and billing for measured use.
    The Internet Society is working on ATM, and carrying Internet packets in 
ATM cells is a mandatory migration tool, sure.  But, the theory of ATM is that 
short fixed-length cells are needed for voice and video applications.  Carrying 
long variable-length packets through an ATM cell-switching fabric is, well, the 
worst of both worlds.  I think the third generation Internet must do better 
than that -- it must exploit ATM with cell-based protocols, operating systems, 
and applications.  Otherwise, the Internet stays stuck in its current 20-year-
old ASCII-bound applications -- TELNET, FTP, and E-mail.
    Another requirement for the third generation Internet is that individuals, 
not just institutions, be allowed to subscribe.  TV and telephone systems are 
driven by personal use, but not yet the Internet.  The Society should be sure 
it's new protocols support personal use, and they should exploit ISDN.  ISDN 
won't replace high-speed Internet trunks, but will greatly improve personal 
Internet use from small offices and home offices -- SOHO networking.  And ISDN 
does something else that the Internet needs, namely measured usage billing.
    The Internet, intended for free use by the scientific community, has no 
billing capabilities.  In fact, a rich set of arguments have been developed in 
the Internet community about why usage billing is really not needed.  For 
example, everyone should be entitled to use the Internet free.  Or, for another 
example, the costs of counting packets are comparable to the costs of carrying 
them and should therefore be avoided.  I'm sorry, but no.
    To go commercial, measured usage billing is essential.  Price is the time-
tested coordinator of supply and demand.  Internet carriers must be able, as 
are telephone companies, to settle with one another for traffic carried on 
behalf of each other's customers.  And end-user billing, as offered by ISDN, is 
needed.  A commercial Internet must be able to bill for usage by kilopackets 
and kilometers -- say in mega-packet-meters (a unit that I yearn to have named 
after me when I pass on).  And if we still want our government to pay the bills 
for certain classes of Internet users, then it can.
    Speaking of money, if you want to join the Internet Society, as I just did, 
to keep in touch with how the third generation is coming along, it costs $70 
per year and gets you the quarterly /Internet Society News/.  Call 703-620-8990.


END


-----------------