Report of the Cannes meeting

Sakae OKUBO okubo at GITI.OR.JP
Sat Jun 13 11:47:02 EDT 1998

Dale L. Skran wrote:
> Sec. 3.4 - Flow Control when using UDP
> The formal liaison comments on maxBitRate. This represents the maximum
> (peak) fax bit rate that the unit can absorb, and should not be
> exceeded over a period of time.

maxBitRate is an extremal-bound peak-rate traffic descriptor which
provides a crude form of open-loop flow control. It is really only
useful across networks where traffic is very smooth and predictable.
Across a small corporate WAN, maybe; the Internet, no. :-) It should
therefore be set to as large a value as is sensible, based on what the
endpoint knows of the connection, which is usually just the speed of the
first segment, if it even knows that much.

> It is interesting that you are
> concerned about UDP flow control as this is a major issue with UDP.
> TCP can be relied on to throttle a high-speed internet fax device to
> match a 4.8 kbps G3 device, but with UDP it is unclear how this will
> happen.

Here's an interesting paper by Jamshid Mahdavi and Sally Floyd on UDP
rate control:
TCP-Friendly Unicast Rate-Based Flow Control

> Compliant terminals should not need a UDP throttle.

I don't understand this. Are you assuming that the gatekeeper will
perform explicit network-state measurement and rate control using ACF
and BRQ, therefore taking the responsibility for UDP flow control
completely away from the endpoint, or are you referring to some kind of
currently nonexistent message that directly throttles the transmitter (I
haven't seen the Liaison report)? Otherwise, assuming we're talking
about an RTP/RTCP stream here, RTCP and a few other measurements provide
sufficient feedback for an endpoint to perform implicit measurement and
control of its transmission rate.

Paul Long___________________________
Smith Micro Software, Inc.__________
"If you are not happy here and now, you never will be." Taisen Deshimaru

More information about the sg16-avd mailing list