Folks, At present we don't track if we lose a burst of packets in a row which means we get a series of loss intervals with length 1. What this means is that performance suddenly slumps. Not good. This fix made a big difference for me - increased performance from 10 Kbits to 200 Kbits on a constrained 1 Mbits link (and then I fixed to higher again later - see my blog). I've got a fix here for this in a basic form. It doesn't address the requirement that they be within the same RTT (measured by CCVal) but is still better than what we have. I realise that this is slightly different to how Gerrit is tackling as he is adding CCVal support and looking at using array. Consider this a "March" patch unless you want to look at sooner. If nothing else it might give ideas for people on how to do it even better. Ian - To unsubscribe from this list: send the line "unsubscribe dccp" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
