The Great Exchange

Tom Walton twalton at dimension.net
Wed May 27 19:30:46 UTC 1998


You're assuming that the internet will persist in its current form, and just
"scale up". I'd argue that, as network technology becomes useful for
non-nerds, consumers will pay for packaged data services and not just for
"access" to an expanded version of today's web-morass. These information
services, e.g., entertainment, news, etc., will be big bandwidth burners.
Simple economics will drive the providers of these high-bandwidth services
to the edges of the network, and hence to a regional model for content
distribution. Thus, the majority of bits, if not the majority of net
"accesses" (one ten minute video stream will likely dominate a residential
consumer's daily consumption) reaching consumers will be delivered
regionally.

You can see the beginnings of this trend in the proliferation of www-caching
research, products, and services (e.g.,  the nlanr caching studies, Inktomi
and Skycache). ISPs are already finding it economical to set up ad hoc
"tiered" content distribution schemes to reduce their backbone bandwidth
requirements. Under this model, the ISPs are themselves regionalizing
content distribution;  what was global traffic, once cached,  becomes local.

As the data services industry matures, and exands to serve the enormous
residential consumer market, large volume content providers will follow this
model, locating servers at regional sites to minimize their backbone costs.

Of course, this blue-sky-ish sort of argument depends on the widespread
deployment of high-speed last-mile access technologies, and the development
of worthwhile services offerings. This will occur at different paces in
different locales; yet its hard (for me, at least) to see a much different
outcome in the long run.

Tom
twalton at dimension.net



>
>"Pickett, David" writes:
>> 1) More needs to be done to leverage locality of traffic
>
>In the long run, why are we assuming there will be locality of
>traffic?
>
>It is true that the old PSTN has locality of traffic, but it doesn't
>have flat rate pricing, or the usage patterns that the Internet has. I
>argue that users are rarely more likely to be trying to download a web
>page from near to their homes than from far away. If there is
>locality, it is probably weak, and in the long run would only account
>for a fraction of the traffic.
>
>Perry




More information about the NANOG mailing list