[R-C] MulTFRC

[Michael Welzl]

Hi,

Since the TCP equation seems to be used in several places by this  
effort and there was some discussion about TFRC today, I thought I  
should remind people of the existence of MulTFRC. We have made an  
equation that gives you the throughput of N flows (where N can also be  
0.3 or 2.9 or whatever...), and we plugged that into TFRC, leading to  
what we call MulTFRC. This has been published in a SIGCOMM poster, ACM  
CCR, a Ph.D. thesis, and now also in IEEE/ACM ToN. Everything about  
MulTFRC is collected here:
http://heim.ifi.uio.no/~michawe/research/projects/multfrc/index.html

The draft:
http://tools.ietf.org/html/draft-irtf-iccrg-multfrc
has been reviewed by 3 reviewers, who all gave their okay, except for  
one reviewer who (quite rightfully!) insisted on us developing a fixed- 
point version of the mechanism. This has turned out to be much harder  
than we thought, and the draft has been dormant since. We're still on  
the case, and hope to be able to update it soon.

MulTFRC doesn't make TFRC any more delay-friendly and hence probably  
won't solve any of the problems that people now seems to have with  
TFRC, but I felt that it could be worthwhile for this crowd to know  
that this equation, and this mechanism, exist.

Cheers,
Michael