RocketMan, from your reply # 2491:
>>with the increasing use of streaming audio/video and e-commerce, not to mention web cam pictures of people's cats. <<
I think that these will have a more profound effect next year, or later, as most users (who I speak with, in any event) still haven't gotten into this mold to the point where this would make a difference at this time. Some great reporting, btw.
From this one [#2494], wherein you quote from the congressional record:
>>Fewer than 40 million people around the world were connected to the Internet during 1996. By the end of 1997, more than 100 million people were using the Internet. As of December 1996, about 627,000 Internet domain names had been registered. By the end of 1997, the number of domain names more than doubled to reach 1.5 million. Traffic on the Internet has been doubling every 100 days. Madam President, I feel compelled to repeat that. Traffic on the Internet has been doubling every 100 days."<<
The problem that I have with this is that the Pareto Effect is most definitely a prevalent factor here to be reckoned with, where, say, 20 % of all web sites, or users, slice it as you may, are accountable for 80% of the traffic. Therefore, I would suggest that to linearly extrapolate across the whole is an invalid assumption that leads, consequently, to an invalid conclusion.
I would further suggest that the VPN and extranet factors play into a part of this very heavily as the 20% or 30% Paretos [ but not as heavily perhaps as some of the portals]. But like I've stated before, the accounting of the VPN segments of traffic by the ISPs are not trustworthy, as reporting is non-uniform, and there are fuzzy areas of distinction between these VPNs/intranets and public-Internet resources, because, in fact, they actually share common networking resources in many cases.
Take the largest private intranets and VPNs that are outsourced by GTE Internetworking, for example, or by Concentric Networks, or any other large ISP. These quasi-private offerings use the same routers and other backbone elements as their public offerings. It would be easy for me to see how they could count the traffic that is handled on the private portion as part of their overall public Internet traffic, because they share the same routers, the same pipes, and in most cases, they share the same accounting and network management systems on a partitioned basis.
Are VPNs and outsourced intranets by ISPs on the rise? Should they be counted in the public Internet traffic being measured here? Are they, in fact?
It would be easy to omit, or to double-count [let's not start with the double counting here, not just yet] some of these quantities, and their associated throughput levels, depending on what argument one was pursuing. Since these will be among the largest users, their significance would be felt more profoundly, in any event. I.e., not proportionally to the other users in the field, but much greater.
On the other had, if there are many multiples of mom and pop domain names, and additional residential hosts being introduced daily, the significance of some of these need not be felt at all, in relative terms. Another reason to dismiss the validity of some of the information that was presented above.
All of this is just IMHO, of course, and comments are welcome. Be back in a couple of days. This time I mean it, and good night, Folks.
Regards, Frank C. |
