Yes! The Internet is on the verge of collapse

Bradner's view

Predicting the Internet's catastrophic collapse and ghost sites galore in 1996 - Metcalfe's original end-of-Net column.

Metcalfe invented Ethernet in 1973 and founded 3Com Corp. in 1979. He receives E-mail at bob_metcalfe @infoworld.com.

The Internet might possibly escape a ''gigalapse'' this year. If so, I'll be eating columns at the World-Wide Web Conference in April. Even so, Scott Bradner should be concerned about the Internet's coming catastrophic collapses.

Collapses are widespread and prolonged Internet outages, which, like tropical storms when catastrophic enough, get named and tracked.

To size an outage, multiply the number of users times their hours of denied access. A recent BBN Planet ''kilolapse'' meant thousands of lost user hours. An ampersand mistyped into a router de-'Netted 400,000 Netcom users for 13 hours - a 5.2 ''megalapse.'' Another botched router update de-Webbed 6.2 million America Online users for 19 hours - a catastrophic 118 megalapse.

Now don't you be confusing megalapses with the bogging down of the Internet. Members of the World-Wide Wait Watchers Club do this. They complain about waiting too long for downloads, too often hitting their ''Stop loading'' buttons and too seldom getting better service by changing Internet service providers. To them, the bogging down and collapses are the same thing - a growing pain.

To work around the bogging and collapsing Internet, many ISPs??? have been building intranets. Even the universities that built the Internet are angling for their very own private ones. When will our thousands of uncooperative ISPs, who count on luring users from their collapsing competitors, learn why airlines don't compete by claiming their airplanes crash less?

Among causes of Internet collapse are traffic jams. Of course, this automobile analogy forgets that when Internet bridges are gridlocked, packets (unlike cars) get dumped into the river. The 10% packet loss that caught my attention last December now sometimes exceeds 40%.

Internet retransmission protocols, when widely implemented, multiply packet losses, slow downloads and often sneak quietly away, leaving inexplicable error messages. There's a real danger here of regenerative retransmission collapse.

Overbuilding the Internet to overcome its architectural problems is decreasingly an option. Financial incentives for adding capacity are weak - ISPs struggle to make money. Despite all that dark fiber, actual circuits are harder to come by. Installation delays on 4M bit/sec backbone circuits, for example, are now stretching out toward 180 days.

Let's be concerned that large portions of the Internet might be brought down, not by nuclear war, but by power failures, telephone outages, overloaded domain name servers, bugs in stressed router software, human errors in maintaining routing tables and sabotage, to name a few vulnerabilities in our beloved house of cards.

Because the Internet's early builders believed that it defies management - It's alive! - they punted, leaving no organized process for managing Internet operations. Where, for example, are circuits inventoried, traffic forecasts consolidated, outages reported, and upgrades analyzed and coordinated? The Internet Engineering and Planning Group and the North American Network Operators Group are by most accounts no-ops (For those of you not up on programming lingo, a no-op is an operation that does nothing).

But the Internet is not alive. It's actually a network of computers. And somebody - hopefully cooperating ISPs - should be managing its operations.

The Internet Engineering Task Force now seems to be coming around. It just reorganized itself to create an Operations and Management (O&M) Area. Scott Bradner should be reelected to lead O&M and to organize ISP processes for cooperative operational management.

Now, some say I'm a control freak, that we should let competition in the free market save us from Internet catastrophe. Excuse me, but pundits whining their concerns, buyers getting informed and competing service providers better organizing their cooperations are all good examples of exactly how this free market thing works. If we get ISPs cooperating on operations soon, maybe our growing governments and their telephone monopolies won't need to step in to save the Internet's day.