RE: Flow label and its uses

["Bora Akyol" <bora@broadcom.com>]

Thanks Brian,

If this field is indeed not commonly used (presumably due to lack
of an API to take an advantage of it in end point OSs),
is it worth it to define it as "Reserved."

I don't want to raise a controversial issue here, but
this 'wild card' field presents some challenges in implementation
of v6 support without well understood applications.

Regards,

Bora
 

> -----Original Message-----
> From: Brian E Carpenter [mailto:brc@zurich.ibm.com] 
> Sent: Tuesday, January 17, 2006 7:44 AM
> To: David Malone
> Cc: Bora Akyol; v6ops@ops.ietf.org
> Subject: Re: Flow label and its uses
> 
> I've been a little disappointed not to see multiple drafts 
> proposing use cases within the rules set by RFC 3697. That 
> document was intended to set the baseline for flow label 
> based classifiers, but use cases are needed too.
> 
>     Brian
> 
> David Malone wrote:
> > On Mon, Jan 16, 2006 at 04:41:53PM -0800, Bora Akyol wrote:
> > 
> >>I am looking at pointers on how commonspread the usage of the "Flow 
> >>Label" field is in IPv6 (specifically end-node operating systems).
> > 
> > 
> > Orla McGann did some measurements of this in her thesis:
> > 
> > 	http://www.redbrick.dcu.ie/~orly/thesis.pdf
> > 
> > Section 5.2.4 (page 101 of the PDF, numbered 91) has some live 
> > measurements of how many hosts were setting the flow label and some 
> > fingerprinting to determine what OS they were running.
> > 
> > 	David.
> > 
> 
> 
>