Re: Flow label and its uses

[Brian E Carpenter <brc@zurich.ibm.com>]

Bora Akyol wrote:
> > From a switching hardware perspective, it would be nice
> to either define the use of this field as --endpoint only--
> or label it "Reserved."

I disagree pretty strongly. As defined now, this field *needs*
to be covered by hardware classifiers. It was to clarify that
point (which is ambiguous in RFC 2460) that the IPv6 WG
produced RFC 3697.
> There has been significant time since RFC3697 and the lack
> of applications may indicate that this field (with
> the exception of NIMROD) may not have a use at all.

NIMROD is ancient history. I firmly believe the use cases
will come, once IPv6 is deployed to the extent that people
really care about load balancing etc.

   Brian
> Regards
>
> Bora
>
>
>
> > -----Original Message-----
> > From: Brian E Carpenter [mailto:brc@zurich.ibm.com] 
> > Sent: Wednesday, January 18, 2006 5:03 AM
> > To: Fred Baker
> > Cc: Bora Akyol; v6ops@ops.ietf.org
> > Subject: Re: Flow label and its uses
> >
> > Fred Baker wrote:
> >
> > > personally, I would label it reserved. I think the authors of RFC  
> > > 3697 see it as something akin to a 20 bit DSCP, and if 
> >
> > someone wants  
> >
> > > to see it that way it's fine by me. In any event, that is 
> >
> > an ipv6wg  
> >
> > > question more than a v6ops question.
> >
> > Indeed. And it isn't "reserved"; it *is* part of the header 
> > with the semantics defined to the extent of RFC 3697. My 
> > point is that people who want to use it, e.g. for load 
> > balancing, ought to be writing drafts.
> >
> >    Brian
> >
> >
> > > On Jan 17, 2006, at 3:50 PM, Bora Akyol wrote:
> > >
> > >
> > > > A pointer to some reference material on NIMROD:
> > > >
> > > > http://ana-3.lcs.mit.edu/~jnc/nimrod/nimsl.html
> > > >
> > > > http://ana-3.lcs.mit.edu/~jnc/nimrod/docs.html
> > > >
> > > > http://www.ir.bbn.com/projects/nimrod
> > > >
> > > > Is there any use in keeping flow label as is, or should be 
> >
> > relabeled 
> >
> > > > as "Reserved"?
> > > >
> > > > I think there is some agreement that
> > > > the label in
> > > > (Label, IP Source Address, IP Destination Address) triplet 
> >
> > does not 
> >
> > > > add a whole lot of value.
> > > >
> > > > Regards,
> > > >
> > > > Bora
> > > >
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: Fred Baker [mailto:fred@cisco.com]
> > > > > Sent: Tuesday, January 17, 2006 3:33 PM
> > > > > To: Bora Akyol
> > > > > Cc: Vishwas Manral; v6ops@ops.ietf.org
> > > > > Subject: Re: Flow label and its uses
> > > > >
> > > > > I'd encourage you to look at the big-internet archives (if they
> > > > > exist) from about 1993. The flow label was proposed to 
> >
> > support the 
> >
> > > > > nimrod architecture, and in essence *was* what we later 
> >
> > described as 
> >
> > > > > "MPLS", but in the IPv6 header. That's one of the reasons 
> >
> > that the 
> >
> > > > > flow label isn't covered by the IPSEC checksum - so it could be 
> > > > > managed appropriately at ingress and egress to the 
> >
> > various "flows" 
> >
> > > > > or "LSPs".
> > > > >
> > > > > Yes, there has been a lot of water under that bridge. Between 
> > > > > requiring the flow label to pass unchanged and making the address 
> > > > > fixed length and of the same construction as the IPv4 address, 
> > > > > Nimrod became very difficult to implement in IPv6, and Noel still 
> > > > > isn't very happy with the IPv6 community.
> > > > >
> > > > > On Jan 17, 2006, at 3:11 PM, Bora Akyol wrote:
> > > > >
> > > > >
> > > > > > > -----Original Message-----
> > > > > > > From: Vishwas Manral [mailto:Vishwas@sinett.com]
> > > > > >
> > > > > >
> > > > > > > And a more recent draft
> > > > > > > http://www.faqs.org/ftp/pub/internet-drafts/draft-chakravorty-
> > > > > > > bcc-flowla
> > > > > > > bel-00.txt
> > > > > >
> > > > > >
> > > > > > This last one looks a lot like MPLS in IPv6 ;-)
> > > > > >
> > > > > > Bora