RE: VPN Identifiers

[Giles Heron <[e-mail]>]

Not sure I'd go with that consensus.

OUIs aren't exactly cheap, are they?

Giles

On Wed, 2003-08-13 at 06:22, [e-mail] wrote:
> Hamid,
> 
> I understand that your GID draft is not intended to solve the problem of
> defining a unique VPN-ID format and I agree that RFC2685 meets the
> requirements for a VPN-ID. If there is consensus that this is the best VPN
> ID to use then I would like to see all the drafts support its use.
> 
> Richard
> 
> > -----Original Message-----
> > From: Hamid Ould-Brahim [mailto:[e-mail]]
> > Sent: 12 August 2003 19:16
> > To: Spencer,R,Richard,XGH5 R; [e-mail]
> > Cc: [e-mail]
> > Subject: RE: VPN Identifiers 
> > 
> > 
> > Richard,
> > 
> > [clipped...]
> > 
> > > 
> > > The point is that currently different mechanisms use 
> > different VPN ID
> > > formats. 
> > 
> > I would say instead "different mechanisms use different global unique
> > identifiers some of them are intended to uniquely identify a VPN."
> > 
> > >If it is decided that 
> > > draft-ouldbrahim-ppvpn-gid-03.txt is the best
> > > VPN ID format to use then that's fine, but for inter-dependency and
> > > interoperability reasons it should be used across all VPN 
> > mechanisms.
> > > Looking at draft-ouldbrahim-ppvpn-gid-03.txt, do we really 
> > > need 4+ different
> > > naming methods for a VPN? I think two options for global 
> > > identifiers at the
> > > most should be sufficient, i.e. AS#, or OUI, although 
> > > obviously one global
> > > ID would be preferable.
> > 
> > draft-ouldbrahim-ppvpn-gid-03.txt tries to define formats for 
> > supporting global unique identifiers mainly intended for VPN 
> > solutions and mostly targeting control planes, and among these 
> > formats it supports the route target and RFC2685 formats within
> > an 8 octets length field. These ids are required to be unique 
> > only within a given solution and they do not necessarily identify 
> > a VPN. A particular solution may decide to assign a gid to uniquely 
> > identify a VPN if it wants to, and may want to use gids for other
> > purposes as well. 
> > 
> > The draft does not solve the problem of defining a unique VPN-ID 
> > format intended to be used everywhere across all solutions including 
> > management. In my view if one wants to solve this problem and enforce 
> > that id on all solutions and on all the components of a given solution
> > then there is already rfc2685 standard which pretty 
> > much meets the definition you mentioned below. Whether solutions 
> > within l2vpns will be using it everywhere is obviously 
> > another thing. 
> > 
> > Hamid.
> > 
> > > 
> > > Richard
> > > 
> > > > -----Original Message-----
> > > > From: Eric Rosen [mailto:[e-mail]]
> > > > Sent: 11 August 2003 18:26
> > > > To: Spencer,R,Richard,XGH5 R
> > > > Cc: [e-mail]; [e-mail]
> > > > Subject: Re: VPN Identifiers 
> > > > 
> > > > 
> > > > 
> > > > We already  have an  8-byte VPN-Id format  which can  be 
> > > > based either  on AS
> > > > numbers or on OUIs, and is  extensible to other possibilities 
> > > > as well.  Have
> > > > you looked at draft-ouldbrahim-ppvpn-gid-03.txt? 
> > > > 
> > > 
> > > 
> > 
> 
> 
-- 
=================================================================
Giles Heron    Principal Network Architect    PacketExchange Ltd.
ph: [phone]              "if you build it they will yawn"
=================================================================