argus and multiple links

[Carter Bullard]

Hey Mark,
   There are some strict round-robin schedulers, but
with the equipment you mention, you will at least get
each half-duplex flow mapped to the same argi.  But,
to do some of the TCP retransmission logic, we still
correlate seq numbers from both halfs of the flow, so
there will still be some information that will be lost.

Carter

Carter Bullard
QoSient, LLC
300 E. 56th Street, Suite 18K
New York, New York  10022

carter at qosient.com
Phone +1 212 588-9133
Fax   +1 212 588-9134
http://qosient.com

> -----Original Message-----
> From: Mark Poepping [mailto:poepping at cmu.edu] 
> Sent: Thursday, July 19, 2001 11:07 AM
> To: carter at qosient.com; 'David Richardson'
> Cc: argus-info at lists.andrew.cmu.edu
> Subject: RE: argus and multiple links
> 
> 
> 
> >    Either loss or retransmissions can be hard to recover. 
> Argus does a 
> > lot of work to try to understand if there are TCP retransmissions.  
> > Transmissions of ACK's are detectable based on "I've 
> already seen this 
> > before", and if your round robining TCP data to 10 different argi, 
> > then the odds of you picking up ACK retransmissions is 
> going to fail.
> 
> Do you really mean round-robin to a bunch of argi, or just 
> monintoring multiple, balanced links along a single path? Do 
> any of the balancing methods split a single flow across 
> multiple paths?  I suppose it can be done, but from what I've 
> seen of OSPF and cisco etherchannel, specific flows generally 
> follow the same path, so the retransmit foo will work (as 
> long as you don't have links flapping).
> 
> mark.
> 
>