RE: acks and pacing

Neal Cardwell (cardwell@cs.washington.edu)
Thu, 24 Jun 1999 01:16:52 -0700 (PDT)

> Correct, however because TCP increments TCP for each ACK, a ACK-n policy
> will make cwnd will grow proportionately more slowly (don't want to do this
> over long paths).

good point.

> As an alternative, a nice switch would be to increase cwnd according to
> bytes acked (not ACK's received) and use pacing to smooth out the bytes
> without that window.

that sounds like a good idea. as another alternative, using vegas should
achieve this as a side effect of the once-per-RTT time scale of vegas
cwnd decisions.

neal