26Feb/1210

6PE

The goal of this exercise is to study 6PE (IPv6 Provider Edge) transition mechanism described in draft-ooms-v6ops-bgp-tunnel-00.txt (now published as RFC 4798, since the draft is very old, the solution proposed here are not exactly what have been standardized). This principle is implemented by many major router manufacturers. This proposal combines one IGP, MPLS and BGP to allow an ISP to carry efficiently IPv6 traffic without modifying core routers. Only edge routers have to be modified to understand IPv6 protocol. In the rest of this exercise we will study progressively the Internet Draft and the different elements of this proposal.

Question 1 Some manufacturers propose to implement IPv6 forwarding functionality in a supervision card. Why this solution is not efficient ?

INTERNET DRAFT               J. De Clercq, G. Gastaud, D. Ooms
<draft-ooms-v6ops-bgp-tunnel-00.txt>          Alcatel
                              S. Prevost
                              BTexact
                              F. Le Faucheur
                               Cisco
                                                                 October, 2002               Expires April, 2003
      Connecting IPv6 Islands across IPv4 Clouds with BGP
[...]

Question 2 What is this document status ? Has it been adopted by an IETF working group ?

Abstract

This document explains how to interconnect IPv6 islands over an IPv4 cloud, including the exchange of IPv6 reachability information using BGP. Two approaches will be explained, both requiring a Dual Stack MP-BGP-speaking edge router per IPv6 island. The hosts in the IPv6 islands can use native IPv6 addresses.

The first approach uses MP-BGP over IPv4, relies on identification of the MP-BGP-speaking edge routers by their IPv4 address and uses a trivial tunneling mechanism without any explicit tunnel configuration. The second approach uses MP-BGP over IPv6 and relies on existing ngtrans tunneling mechanisms to tunnel packets.

1. Introduction

This document explains how to interconnect IPv6 islands over an IPv4 cloud, including the exchange of IPv6 reachability information using BGP. Two approaches will be explained, both requiring a Dual Stack MP-BGP-speaking edge router per IPv6 island. The hosts in the IPv6 islands can use native IPv6 addresses.

The first approach uses MP-BGP over IPv4, relies on identification of the MP-BGP-speaking edge routers by their IPv4 address and uses a trivial tunneling mechanism without any explicit tunnel configuration. The second approach uses MP-BGP over IPv6 and relies
on existing ngtrans tunneling mechanisms to tunnel packets.
[...]

2. Terminology

[...]
In this document an 'IPv6 island' is an IPv6-upgraded network (which can be cross-AS). A typical example of one island would be one or more Customer IPv6 sites connected via their Customer Edge (CE) router to one (or more) Dual Stack Provider Edge (PE) router(s) of a Service Provider.

3. Applicability

The transition methods described in this document typically applies to an ISP that is familiar with BGP [...] and that wants to offer IPv6 services to some
of its customers. However, the ISP does not (yet) want to upgrade its network core to IPv6. With the mechanisms described here, the provider only has to upgrade some PE routers in some POPs to Dual Stack MP-BGP routers. The Dual Stack MP-BGP routers provide access to IPv6 customers and may provide access to IPv4 customers in addition.


Question 3 AS A want to exchange IPv6 traffic with AS-C. Indicate in the following scheme which routers has to be modified.

The ISP may also have access to the global IPv6 Internet. The ISP provides global IPv6 connectivity through its peering relationship with an upstream ISP, or by peering relationships with other IPv6 ISPs in the default free routing zone (DFZ).

Question 4 What is a DFZ ?Is AS-X necessarely a DFZ ?

A Dual Stack MP-BGP router in the provider's network is connected to an upstream IPv6 ISP or forms part of the IPv6 backbone network, such as the 6bone. The ISP advertises IPv6 reachability of its IPv6 allocated prefix using MP-BGP to its IPv6 upstream provider or into the IPv6 DFZ. The IPv6 prefixes received from the upstream provider or from the DFZ can be redistributed within the ISP using MP-BGP.

The interface between the CE router and the PE router is a native IPv6 interface which can be physical or logical. A routing protocol (IGP or EGP) may run between the CE router and the PE router for a customer IPv6 site to exchange its reachability. Alternatively, static routes and/or a default route may be used on PE and CE to control reachability. A customer site may connect to the provider network over more than one interface.

Question 5 For AS-X topology shown before, what is the best solution ? justify your answer (IGP, static route, default route, …) ?

The methods in this document can be used for customers that already have an IPv4 service from the network provider and additionally require an IPv6 service, as well as for customers that require only IPv6 connectivity. In both cases the network provider allocates IPv6 addresses to the site.

4. Description

Each IPv6 site is connected to at least one Dual Stack MP-BGP speaking edge router that is located on the border with the IPv4 cloud. We refer to such a router as a DS-BGP router. The DS-BGP router MUST have at least one IPv4 address on the IPv4 side and one IPv6 address on the IPv6 side. The IPv4 address MUST be routable in the IPv4 cloud.

We refer to the DS-BGP router receiving IPv6 packets from an IPv6 site as an Ingress DS-BGP router (relative to these IPv6 packets). We refer to a DS-BGP router sending IPv6 packets to an IPv6 site as an Egress DS-BGP router (relative to these IPv6 packets).

Question 6 On the following scheme, indicate eBGP and iBGP peerings, if we consider that only AS A and AS C are interested by IPv6 traffic. 

Question 7 With which IP version the iBGP TCP connection will be established ?

No extra routes will be injected in the IPv4 cloud.

Interconnecting IPv6 islands over an IPv4 cloud requires following steps:

(1) Exchange IPv6 reachability information among DS-BGP Routers:

(1.a) The DS-BGP routers exchange, via MP-BGP [MP-BGP], IPv6 reachability information over the IPv4 cloud with their peers.

(1.b) In doing so, the Egress DS-BGP routers announce themselves as the BGP Next Hop.

(2) Tunnel IPv6 packets from Ingress DS-BGP Router to Egress DS-BGP Router: the Ingress DS-BGP router tunnels an IPv6 packet over the IPv4 cloud towards the Egress DS-BGP router identified as the BGP Next Hop in step (1.b) for the packet's destination IPv6 address.

We distinguish two approaches for connecting IPv6 islands across IPv4 clouds via BGP, which are respectively referred to as the "MP-BGP over IPv4" approach and the "MP-BGP over IPv6" approach.

Step (1.a) is identical for both approaches.

Steps (1.b) and (2) differ between the two approaches.

4.1. "MP-BGP over IPv4" approach

With this approach, the DS-BGP routers MUST run MP-BGP over an IPv4 stack (MP-BGP/TCP/IPv4). The DS-BGP router conveys to its peer its IPv4 address as the BGP Next Hop.

Since MP-BGP requires that the BGP Next Hop is of the same address family as the NLRI, this IPv4 address needs to be embedded in an IPv6 format. The IPv4-mapped IPv6 address is defined in [V6ADDR] as an "address type used to represent the addresses of IPv4 nodes as IPv6 addresses", thus this precisely fits for the above purpose. Encoding the routable IPv4 address into a IPv4-mapped IPv6 address allows the remote DS-BGP router to automatically tunnel data over the IPv4 cloud to the destination IPv6 island.

Question 8 AS-C announces the IPv6 prefix to AS-X. Internally what will be the value of the NEXT-HOP  field announced to router A?

Any type of encapsulation can be used (IPv4, MPLS, ...).

Question 9 Why MPLS allow to carry IPv6 packets, even if core routers are not been config protocol version ?

In the "MP-BGP over IPv4" approach the IPv4 address of the MP-BGP next hop MUST be encoded as an IPv4-mapped IPv6 address.

The ingress DS-BGP Router MUST tunnel IPv6 data over the IPv4 cloud towards the Egress DS-BGP router identified by the IPv4 address advertised in the IPv4-mapped IPv6 address of the BGP Next Hop for the corresponding IPv6 prefix.

[…]

5. Tunneling

5.1. "MP-BGP over IPv4" approach

In the "MP-BGP over IPv4" approach, the IPv4-mapped IPv6 addresses allow a DS-BGP router that has to forward a packet to automatically determine the IPv4 endpoint of the tunnel by looking at the MP-BGP routing info.

If this approach is used to connect to the public IPv6 Internet, tunnels without special security mechanisms MAY be used (e.g. IPv4 tunnels [TUNNEL], GRE tunnels [GRE] or MPLS LSPs without IPSec).

Note that even when the number of peers is high, the number of  tunnels is not a scalability concern from an operational viewpoint since those are automatic tunnels and thus require no configuration.

Considerations on 'common tunneling techniques' in [TRANS] are valid for this approach.

[…]

5.1.2. Tunneling over MPLS LSPs

When the IPv4 backbone supports MPLS, MPLS LSPs MAY be used as the tunneling technique. These LSPs can be established using any existing technique (LDP, RSVP, ...).

When MPLS LSPs are used with the "MP-BGP over IPv4" approach, rather than successively prepend an IPv4 header and then perform label imposition based on the IPv4 header, the ingress DS-BGP Router MAY directly perform label imposition of the IPv6 header without prepending any IPv4 header. The (outer) label imposed corresponds to an LSP starting on the ingress DS-BGP Router and ending on the egress DS-BGP Router.

While the "MP-BGP over IPv4" approach can operate using a single level of labels, there are advantages in using a second level of labels which are bound to IPv6 prefixes via MP-BGP advertisements in accordance with [LABEL]. For instance, use of a second level label allows Penultimate Hop Popping (PHP) on the Label Switch Router (LSR) upstream of the egress DS-BGP router without any IPv6 capabilities/upgrade on the penultimate router even when the IPv6 packet is directly encapsulated in MPLS (without an IPv4 header); since it still transmits MPLS packets even after the PHP (instead of having to transmit IPv6 packets and encapsulate them appropriately). Thus, the "MP-BGP over IPv4" approach MAY be used with a single label and MAY also be used with a second label.

Question 10 On the following drawning give an example of FEC, switching table and MPLS (encapsulating IPv6 packets) that will be send between A and C LERs?

This content is published under the Attribution-Noncommercial-No Derivative Works 3.0 Unported license.

Tags: , , , , ,

10 Responses to “6PE”

  1. admin says:

    Question 1 Some manufacturers propose to implement IPv6 forwarding functionality in a supervision card. Why this solution is not efficient ?


    Forwarding in the supervision card means that the router can process only one by one packets, it is not possible to forward simultaneously several IPv6 packets as with network cards.
  2. admin says:

    Question 2 What is this document status ? Has it been adopted by an IETF working group ?

    This is a personal draft, not endorsed by a working group (even if it focuses v6ops) so it was not at that time adopted by an IETF working group.
  3. admin says:

    Question 3 AS A want to exchange IPv6 traffic with AS-C. Indicate in the following scheme which routers has to be modified.


    Only PE A and PE C has to take into account IPv6, the other devices do not need to be modified
  4. admin says:

    Question 4 What is a DFZ ?Is AS-X necessarely a DFZ ?


    DFZ means Default Free Zone. It is the part of the internet where routers do not have a default route in their routing tables (they need to know all the prefixes (aggregated) used on the Internet). AS-X can not be in a DFZ, 6PE can be used in an entreprise network.
  5. admin says:

    Question 5 For AS-X topology shown before, what is the best solution ? justify your answer (IGP, static route, default route, …) ?


    BGP is the simplest way to inform AS-X 
  6. admin says:

    Question 6 On the following scheme, indicate eBGP and iBGP peerings, if we consider that only AS A and AS C are interested by IPv6 traffic. 


    eBGP between 
    – AS-A and AS-X, 
    – AS-B and AS-X
    – AS-C and AS-Y

    iBGP between
    routers A and B
    routers A and C
    routers B and C

    iBGP connection are not transitive, so full mesh is needed. This full mesh is independent  of the nature of the exchange. IPv6 capabilities will be negotiated only between router A ebt C where IPv6 NLRI will be exchanged.
  7. admin says:

    Question 7 With which IP version the iBGP TCP connection will be established ?


    IPv4, P router are not aware of IPv6.
  8. admin says:

    Question 8 AS-C announces the IPv6 prefix to AS-X. Internally what will be the value of the NEXT-HOP  field announced to router A?


    ::FFFF:η.2
  9. admin says:

    Question 9 Why MPLS allow to carry IPv6 packets, even if core routers are not been config protocol version ?


    Because switching is done before Layer 3 header so anything can be carried over MPLS.
  10. admin says:

    Question 10 On the following drawning give an example of FEC, switching table and MPLS (encapsulating IPv6 packets) that will be send between A and C LERs?



    router C LER
    prefix from A => L1
    Router C BGP Table
    Prefix from A => L111
    From C to G
    IPv6 
    MPLS L1 / L111
    phys
    Router G LIB
    inter 3:L1 -> inter 1:L2
    From G to F
    IPv6 
    MPLS L2 / L111
    phys
    Router F
    inter 2:L2 -> inter 1:L3
    From F to D
    IPv6 
    MPLS L3 / L111
    phys
    Router D
    Inter 2 L3: pop
    From D to A
    IPv6 
    MPLS  L111
    phys

Leave a Reply

Your email address will not be published. Required fields are marked *