topological "closeness" (RE: Web o' Wonder)

[Mike Trest <trest () atmnet net>]

Michael Dillon's Web o' Wonder scenario keys on the ability of smart servers
to do an application level re-direction of load to a topographically more
appropriate server.  Some of our customer's sites are doing a fair job of
this by weighting DNS responses to geographically dispersed sites which,
hopefully, are topographically dispersed.

> Seems to me that this scenario would also benefit from a tool and database
> that could determine topological "closeness" even if it doesn't need to
> generate filter lists.
> If this scenario were easier to implement it could reduce the load of the
> major exchange points by encouraging traffic to stay closer to the network
> periphery.

The CACHE research efforts (see: http://www.nlanr.net and others) indicates
that even greater reductions would be obtained if NSP/ISPs increase use of
caches for www. Individual campus statistics, the NLANR prototypes, and
statistics from CACHE sites deployed internationally suggest that
orders-of-magnitude reductions in external network access [i.e. our backbone
load] could be achieved.

I am implementing a CACHE simply to improve on-net customer satisfaction.  I
expect the operational costs to be much less than the equivilent backbone or
transit costs. If it has a side effect of reducting meet point traffic, then
the entire net benefits.

I wonder, how many nets would implement such as cache if the effect would be
to reduce meet point traffic by orders-of-magnitude?

..mike..
Mike Trest,  ATMNET          Voice:  619 643-1805
5440 Morehouse Drive         Fax:    619 643-1901
San Diego, CA  92121         Pager:  619 960-9070

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