The following text is copyright 1996 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

Internet health report, a second opinion

A few weeks ago this paper did a two part story, replete with traffic warning signs and yellow and black striped borders concluding that the Internet is not ready for prime time. Its time for a second opinion on that diagnosis.

Aside from some amusing terminology glitches (CIDR is Classless Interdomain Routing, not Complex Interdomain Routing - even if the implications of CIDR can sometimes be complex.) the articles presented generally accurate information but were a bit too tabloid in the interpretation of the information. The tone of the articles and the accompanying graphics implied that only a fool would even consider using the Internet for anything Important let alone actually do so. This seems somewhat out of line with results of the Network World survey reported in the same articles in which 88% of the readers reported that they were satisfied to extremely satisfied with the reliability of the Internet and only 10.5% said they were not satisfied or not very satisfied.

The most basic problem with these articles, as with many other commentaries on the Internet is that they treat the Internet as a monolith. If a user of your web site is getting lousy performance then the Internet must be at fault even if the real problem is the fact that the user is on a 2400 baud dial-up. The specific requirements of a particular Internet user must be understood before deciding that the Internet is worthless. If you wish to reach one of your own branch offices and your company has contracted with a single top of the line ISP for Internet connections to all branches, you are likely to quite happy with the result. I suppose you could be doing things on the cheap and try and support a 50 person office over a 28,8Kb dial up (sadly enough, I have seen just that)but do not blame the ISP or the 'Net for your resulting performance problems.

On the other hand, if you have venture capital money to set up an Internet based phone company and expect to be able to reliably reach customers across multiple ISPs and through multiple exchange points to the ends of the earth, take the money and run - cuz it ain't going to work.

Most organizations that want to use the Internet for business fall between these two extremes. Today a connection to one of the major ISPs will get you quite reliable service to customers of that ISP as well as customers of many of the other major ISPs. These top-line providers are growing high speed direct inter-ISP connections to off load traffic from the normal exchange points. In many cases the only sustained performance problems users will see are related to the speed of their own tail circuit as well as the tail circuit speed and load of the services they wish to use. As above, it is hard to see that the whole 'Net is somehow broken because some company underestimates the demand for information about a human vs. computer chess match and tries to run the web server on a small workstation behind a low speed link to an ISP.

Clearly you can find the kind of problems mentioned in the articles if you work at it -- for example, buy your service from an ISP whose source of clues is a set of Cliff notes and get a link that supports about 10% of your need -- I guess 88% of the survey respondents did not try hard enough.

disclaimer: Although abundantly supplied with opinionated pundits like me, the University itself does not second guess the diagnosis of others.