Hi Don,

Thanks for summarizing issues.

Questions for clarification.

- Does the signaling change for shuffling only? Or for nesting and
  stitching as well?

- If we introduce RD in addition to "L1vpn Globally unique I-D"
  specified in the current I-D, we open the door to the extranet, I
  assume. Does it imply that PIT is per RD, rather than per L1vpn
  globally unique I-D?

- It may be too early to say, but does this change impact how CE-PE TE
  information is distributed (e.g., by BGP-TE)?


Thanks,
Tomonori

On Thu, 1 Mar 2007 14:56:02 -0500
"Don Fedyk" <[EMAIL PROTECTED]> wrote:

> Hi
> 
> An issue came up when we were discussing optimizing the L1VPN auto
> discovery information and also an implication if we move to multiple
> domains. 
> 
> For reference the current the L1VPN basic mode draft has:
> 
>  
>    (Client realm) 
>    +----+                             +----+ 
>    |    |<Port Index>    <Port Index> |    | 
>    |    |CPI              VPN-PPI     |    | 
> ---| CE |-----------------------------| PE |--- 
>    |    |                <Port Index> |    | 
>    |    |                 PPI         |    | 
>    +----+                             +----+   
>                                  (Provider realm) 
> 
> 
> And the auto discovery information is documented as: 
> 
> +---------------------------------------+ 
> |     Length (1 octet)                  | 
> +---------------------------------------+ 
> |     PPI Length (1 octet)              | 
> +---------------------------------------+ 
> |     PPI (variable)                    | 
> +---------------------------------------+ 
> |     CPI AFI (2 octets)                | 
> +---------------------------------------+ 
> |     CPI (length)                      | 
> +---------------------------------------+ 
> |     CPI (variable)                    | 
> +---------------------------------------+ 
> 
> 
> When Signaling the procedure is to look up the remote CPI a the near end
> PE and translates to a PPI for signaling purposes. If we were to extend
> this information to multiple domains the PPI-CPI mapping would carry the
> Provider address PPI outside the provider. 
> 
> One way to optimize this is to carry only the CPI with a Route
> Distinguisher, which is an 8 byte globally unique value. 
> 
> Like this: 
> 
> +---------------------------------------+ 
> |     Length (2 octets)                 | 
> +---------------------------------------+ 
> |     RD (8 octets)                     | 
> ~                                       ~
> +---------------------------------------+ 
> |     CPI1 (length 1 octet)             | 
> +---------------------------------------+ 
> |     CPI1 (variable)                   | 
> +---------------------------------------+ 
> |     CPI2 (length 1 octet)             |
> +---------------------------------------+
> |     CPI2 (variable)                   | 
> +---------------------------------------+
> |     etc                                         |   
> 
> While the RD is larger the actual space for carrying multiple CPIs will
> be reduced. The remote PE is inferred from the BGP Next Hop carried in
> the auto-discovery routes.  The information in these records is all
> globally unique addressing information and independent of provider
> attachment. This aligns with the other auto discovery schemes.
> 
> Note the PPI and VPN-PPI remain but are local information only. 
> 
> This also implies a signaling change.  The changes would include look up
> CPI for a given RD, Signal to the destination associated PE and include
> the RD and CPI in signaling to allow the destination PE to resolve the
> customer port information.  
> 
> We would like to solicit comments on these changes. 
>   
> Thanks,
> Don 
> 
> _______________________________________________
> L1vpn mailing list
> L1vpn@lists.ietf.org
> https://www1.ietf.org/mailman/listinfo/l1vpn


_______________________________________________
L1vpn mailing list
L1vpn@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/l1vpn

Reply via email to