toad wrote: > Okay, so there is a dramatic collapse because it reaches the maximum > capacity of the network. Whereas with flow control, we keep on adding > requests even though the network is overloaded, and they are misrouted, > and not tried properly; we progressively route requests to fewer and > fewer nodes.
With backoff, yes - backoff effectively removes links from the network. With throttling alone there's no misrouting. > So the axes on the graph are a little misleading. We start very many > requests, but we don't properly attempt them. The throughput graph shows the number of successful requests, not the number of attempts. > So throttling should, on your axes, cut out > some time before an unthrottled network cuts out. I disagree - the throttled network should never cut out. It should be able to sustain roughly the same throughput it reached just before the collapse point, no matter how high the load gets. And that's roughly what we see. > We have to design the network on the basis that a lot of users (maybe > most users) will have lots of files queued up; any throttling that is > necessary to manage demand must be executed _by the network_. Agreed, the main advantage of token passing over throttling is that it doesn't rely on users to limit their own usage. > Ideally > this means that the network would have nearly full performance up to > some amount of input load, and then it would just stop, having reached > its input load limit. Not sure what you mean by just stop - beyond a certain load, the throughput should stop increasing and the success rate should stop decreasing. Cheers, Michael
