On Sat, Jun 16, 2001 at 01:11:58AM +0200, Oskar Sandberg wrote:
> I will experiment with variations on the caching, but removing it is not
> what we want.

I think that the first thing we should try is caching with a probability
in proportion to the distance from the datasource (ie. as the DataReply
moves back through the request chain the data is cached with a
decreasing probability).  Oskar has already proposed a way to do this.

Ian.

PGP signature

Reply via email to