RE: acks and pacing
Amit Aggarwal (amit@cs.washington.edu)
Thu, 24 Jun 1999 01:16:22 -0700 (PDT)
>
> 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. Then you just need to ensure that the ACK rate is high
> enough to prevent the sender from going idle (e.g. at least one ack per
> window). You also presumably want to have some min threshold on the ack
> frequency so you have some guaranteed response-time to network changes.
>
This is exactly what is proposed in Hari and Venkat's paper on effects of
path asymmetry on TCP performance. They increase the congestion window
based on data acked and use pacing to prevent burstiness. However, they
used it for environments where the reverse path is bandwidth constrained
and not in the context of high speed networks.
-Amit