Re: [mpls] entropy label indicator labelindraft-kompella-mpls-entropy-label

Kireeti,

There is one technical reason to have one common ELI label value: for
multicast packets. Multicast message may be sent to many egress PEs. Ingress
PE can only put one value for ELI label. So this enforces all PEs use the
same ELI label value. Is this right?

Regards,
Lucy

> -----Original Message-----
> From: Kireeti Kompella [mailto:kireeti@juniper.net]
> Sent: Wednesday, July 28, 2010 6:19 AM
> To: Yong Lucy
> Cc: mpls@ietf.org
> Subject: Re: [mpls] entropy label indicator labelindraft-kompella-mpls-
> entropy-label
> 
> Hi Lucy,
> 
> On Jul 28, 2010, at 04:06 , Yong Lucy wrote:
> 
> > Yes, each egress PE reserves its own ELI label value. Ingress PEs need
> > remember each PE reserved ELI label. For a large network, one PE may
> send
> > packets to many PEs.
> 
> We might be making a bit too much of this.  A PE needs to allocate for
> reception only one ELI for *all* its apps/protocols.  In fact, a vendor
> will probably have a single ELI value across all their boxes.  We (the
> mpls WG) could pick a suggested value (outside the reserved range) that
> several vendors could choose to implement.  The net is that a PE even
> though it sends to many other PEs may have a small handful (or even one)
> value of ELI to send.
> 
> Folks, speak up, and if there is enough support, I'll throw in a para
> suggesting a common ELI value ... unless the chairs/ADs step in and
> provide technical reasons why not.
> 
> Kireeti.
> 
> > Cheers,
> > Lucy
> >
> >> -----Original Message-----
> >> From: mpls-bounces@ietf.org [mailto:mpls-bounces@ietf.org] On Behalf Of
> >> Curtis Villamizar
> >> Sent: Wednesday, July 28, 2010 2:38 AM
> >> To: stbryant@cisco.com
> >> Cc: mpls@ietf.org
> >> Subject: Re: [mpls] entropy label indicator labelindraft-kompella-mpls-
> >> entropy-label
> >>
> >>
> >> In message <4C4EEC02.2040202@cisco.com>
> >> Stewart Bryant writes:
> >>>
> >>> John
> >>>
> >>> Why do you prefer to use an ELI as opposed to using a new set of FECs
> >>> with the property that they are followed by an EL?
> >>>
> >>> Stewart
> >>
> >>
> >> Stewart,
> >>
> >> The same TE LSP can carry an IP payload and other MPLS LSP.  This
> >> avoids confusing an entropy label with a forwarding label.
> >>
> >> The egress does not have to know apriori whether the ingress is adding
> >> an entropy label and the ingress is free to add an entropy label to
> >> some but not all traffic.
> >>
> >> The draft only provides LDP signaling.  TE signaling should also be
> >> provided.  The entropy label can be a per host value (platform label
> >> space) and it doesn't matter much where in OSPF-TE or ISIS-TE the TLV
> >> is placed.
> >>
> >> It would be nice if as we add further MPLS extensions we got the new
> >> behavior we are looking for without making it such that MPLS-TP cannot
> >> be accommodated on the same path and adding signaling to TE as well as
> >> LDP is a small step in that direction.
> >>
> >> Curtis
> >> _______________________________________________
> >> mpls mailing list
> >> mpls@ietf.org
> >> https://www.ietf.org/mailman/listinfo/mpls
> >
> > _______________________________________________
> > mpls mailing list
> > mpls@ietf.org
> > https://www.ietf.org/mailman/listinfo/mpls