On Tue, Oct 07, 2003 at 01:37:24AM +0200, Simon Lodal wrote: > > Hello > > Harald, I read your paper, and I'd love something like this. As I > understand it, you imply a one master / multiple slaves model. > However, what would prevent your protocol from being used in a > master/master setup? the protocol is not strictly limited to that. However, you would need a way of distributing the load between the nodes _without_ a loadsharing box in front of them, because otherwise this box again has the problem of having to deal with all packets. > That would be much better, sometimes one fast machine is not enough. > Each master could have a lighter load, and therefore more ressources > to communicate with the others. And when one master goes down, only a > group of users will be bothered by the delay before the others catch > up it's traffic. This is theoretically an option once the master/multiple slave model has been implemented. But don't underestimate the complexity behind this. > Simon -- - Harald Welte <laforge@xxxxxxxxxxxxx> http://www.netfilter.org/ ============================================================================ "Fragmentation is like classful addressing -- an interesting early architectural error that shows how much experimentation was going on while IP was being designed." -- Paul Vixie
Attachment:
pgp00605.pgp
Description: PGP signature
