On Sat, Jun 27, 1998 at 09:39:26PM -0400, James R. Van Zandt wrote: > Avery Pennarun <[EMAIL PROTECTED]> writes: > >Jason Gunthorpe wrote: > >> My question is how can we build a single weighted score for each > >> site? Let the user pick one of the top two or so. But how do you > >> weight? > > > >Assuming that the packet loss and timing statistics were made more > >reliable... > > > >I would say that packet loss and delay are of almost equal > >importance, and hop count considerably less. > > I suggest: > (1 - packet_loss) / average_delay > > If you multiplied this by the size of the packets, it would be > approximately the bandwidth (bytes per second).
Unfortunately, packet loss causes a disproportionate amount of trouble; if I miss 20% of packets, I don't get 80% of bandwidth. I get maybe 10% of bandwidth because TCP starts fizzling out. Hence my big mess of a math formula assigning packet loss and delay approximately equal weighting. A 50% site with 10ms delay simply should never win compared to a 90% site with 200ms delay. IMHO... best thing to do is take a lot of results, transfer a file from each of the "obviously better" hosts, and see what formula results in the closest result :) Have fun, Avery -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
