[e2e] An alternative algorithm of calculating RTO

[Colin Percival]

At 15:51 03/06/2002 -0700, Christian Huitema wrote:
>In a heavy
>tail distribution, there is a significant possibility to observe a "very
>long" sample; if the only filtering in effect is exponential smoothing,
>then this aberrant sample is going to remain in the averaged results for
>a very long time. The running average procedure that is being proposed
>is not significantly better in that respect -- the spike would remain in
>effect for 8 to 32 RTT.

   A good start would be to discard samples outside [0,SRTT+K*RTTVAR] -- 
ie, extend the "don't measure RTT on retransmitted segments" rule to "don't 
measure RTT on segments which could have been retransmitted even if we 
didn't actually retransmit because our clock is too slow".

Colin Percival