Re: [mpls] [PWE3] ELI as a reserved label

[John E Drake <[e-mail]>]

Comments inline

Sent from my iPhone


> -----Original Message-----
> From: Yong Lucy [mailto:[e-mail]]
> Sent: Sunday, August 08, 2010 2:36 PM
> To: John E Drake
> Cc: 'Shane Amante'; [e-mail]; [e-mail]; [e-mail]
> Subject: RE: [PWE3] [mpls] ELI as a reserved label
> 
> Hi John,
> 
> > -----Original Message-----
> > From: John E Drake [mailto:[e-mail]]
> > Sent: Wednesday, August 04, 2010 8:49 AM
> > To: Yong Lucy
> > Cc: Shane Amante; [e-mail]; [e-mail]; [e-mail]
> > Subject: Re: [PWE3] [mpls] ELI as a reserved label
> >
> > Lucy,
> >
> > I think you are using the term 'large flow' to mean a large number of
> > packets exchanged between a particular source & destination?
> 
> [LY] No, this is not the "large flow" definition. The definition of the
> flow
> in the draft is "A group of packets that contain the same flow
> "identity" in
> their header. The network intends to transport the packets that belong
> to
> the same flow over the same path so flow packet sequence is preserved."
> And
> the large flow definition is "a flow comes to the network at the
> bit/packet
> rate that is on the top bit rate rank among all the flows".

JD:  You actually gave two different definitions.  How do you identify the beginning and end of a large flow, and how do you know it is a large flow before it has ended?  I.e., if you don't know whether it is a large flow, how can you identify it as such?  Given that large flows will have different packet counts, how does a large flow indicator help a transit node perform better load balancing than simply counting the packets it sends along each path?