On Thu, Nov 30, 2006 at 10:30:56AM +0000, Michael Rogers wrote: > Here are some preliminary simulation results where 10% of the nodes have > 5 kbyte/s bandwidth rather than 15 kbyte/s. The most interesting change > is that the collapse point with no flow control has moved from roughly > 11 requests/node/minute to roughly 6.
Are all the lines shown to exhaustion? "No flow control" stops at around 6 because above input load = 6 the network collapses? Why does it collapse? The other lines could be continued? It would be interesting to continue the green, blue and red lines; red seems to be overtaking green at 15, probably this is the result of the expected beneficial effects of backoff (isolating throttling from the effects of really slow nodes) ? It would be useful to see whether red can sustain a lead over green. Is there an endpoint, or does it tend towards a limit on throughput? (Obviously success rate will have to drop eventually). At around 10, the success rate and the throughput start to increase with increasing input load, with throttling, with or without backoff; this is rather surprising too; I'd expect the throughput to increase a bit, but the success rate increasing is surprising. It would appear that with our current load management system, a relatively large failure rate is part of the system, and we need to plan on that basis; this may have design implications, but it's not necessarily catastrophic. > Cheers, > Michael -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20061130/664603f3/attachment.pgp>
