[e2e] High Packet Loss and TCP (TCP throughput)

Taehyun Kim tkim at cc.gatech.edu
Mon May 5 13:19:15 PDT 2003


I think this is a very specific question about the paper.  But when I
followed every step in the paper, I encountered a question: In deriving
the last equation (equation 29), it seems the (1+32p^2) term is
approximated by f(p)/(1-p) where f(p) = 1+p+2p^2+4p^3+...+32p^6.  Hence,
f(p)/(1-p) = 1+2p+4p^2+8p^3+...  But this is significantly larger than
1+32p^2.
Can anybody let me know where the (1+32p^2) term comes from?  Thank you,

tkim


> Exactly.  The paper that Jonathan Smith referred to in the earlier
> message on this thread is:
>   Modeling TCP Throughput: A simple model and its empirical evaluation
> (1998)
>   Jitendra Padhye, Victor Firoiu, Don Towsley, Jim Kurose,
>   http://citeseer.nj.nec.com/padhye98modeling.html
> which points out the importance of timeout when modeling throughput as a
> function of loss.
> 
> Jim
> 
> >One can argue that this can lead to congestion collapse: if too many
> >connections try to use a too small link, then they will all pack-off to
> >this one packet every 4 second behavior, which may well exceed the link
> >capacity. Also, the slow progress will lead to many job aborted midway,
> >which is a loss of resource. It has been shown that, in these
> >circumstances, admission control helps: preferably drop the SYN packets,
> >so that the remaining connections have some bandwidth and can actually
> >complete.
> >
> >-- Christian Huitema
> >
> >

-- 

=============================================================================
Taehyun Kim	
Networking and Telecommunications Group, (Tel) 404-894-6739
Georgia Institute of Technology          (E-mail) tkim at cc.gatech.edu
Atlanta, GA 30332                        (URL)
http://www.cc.gatech.edu/~tkim
=============================================================================




More information about the end2end-interest mailing list