Previous Page
Next Page

Overview

This chapter covers IPv6 unicast connectivity within an enterprise or service provider network, and discusses available options, implementation scenarios, and deployment recommendations. The intent is to present this information as it applies to different parts of the network:

  • Hosts or Customer Edge (CE) routers

  • Access and backbone infrastructures

The concepts that are relevant in making a host operational in an IPv6 network are presented first. This section of the chapter reviews the mechanisms for providing hosts and CE routers with an IPv6 address, and introduces new concept such as prefix delegation, for providing name-resolution support and for some AAA (authentication, authorization and accounting) management.

Subsequent sections discuss the delivery of IPv6 unicast connectivity from a network perspective. The service deployment will most likely interact at some point with or over some segments with existing IPv4 infrastructures. In this interaction, there are three deployment approaches:

  • IPv6 only IPv6 is the only protocol running over a given link (physical or virtual).

  • Dual stack IPv4 and IPv6 run together over the same link (physical or virtual).

  • IPv6 at the edge only IPv6 is confined to sites located at the edges of an existing IPv4 core infrastructure that it has to traverse.

These three approaches define a framework for the service deployment strategy. The technological means to implement them are presented in the context of the two relevant layers of the network: access and backbone.

Table 3-1 summarizes the deployment mechanisms presented in this chapter. Along with the scope and significant features, the table lists the areas of the network where the mechanism would fit best.

Table 3-1. IPv6 Unicast Deployment Mechanisms

Mechanism

Scope

Where Used

Key Features

Limitations

Native IPv6

Routed

Host/Inter-site

Access/Backbone

Simplest method in terms of setup.

Support on all network elements in the path.

 

Bridged (IPv6 RBE)

Host

Access

IPv6 RBE routes the IPv6 traffic out of a bridged ATM encapsulation.

 
 

PPP

Host/Inter-site

Access

  

IPv6 over MPLS

Dedicated links/circuit

Inter-site

Backbone

It is a native method where a physical or virtual link interconnects two sites.

 
 

6PE

Inter-site

Backbone

MPLS infrastructure. IPv4 Label Switch Path setup. IPv6 transport over MPLS.

Applicable to MPLS infrastructure only. IPv4 and IPv6 traffic follow the same paths.

 

IPv6 MPLS

Inter-site

Backbone

MPLS infrastructure. Native IPv6 Label Switch Path setup. IPv6 transport over MPLS

MPLS only. LDP and RSVP IPv6 implementation not available.

Tunnels

Configured

Host/Inter-site

Access/Backbone

Static, supported by most IPv6 implementations.

Tunnel endpoints can be secured with IPv4 IPsec.

Management overhead.

 

Tunnel broker and tunnel server

Host

Access

Provides the means to automatically set up configured tunnels upon request. A dedicated device or a router performs this function.

Potential security implications.

 

Teredo

Host

Access

Works through multiple NATs.

Management overhead.

 

GRE

Inter-site

Access/Backbone

Static. It supports the transport of layer 2 multicast, which is important for the control plane of some protocols (IS-IS).

Management overhead.

 

ISATAP

Host/Inter-site

Access/Backbone

Automatic tunnel.

Performance. No solution for multicast.

 

6 to 4

Inter-site

Access/Backbone

Automatic tunnel. Reserved address space 2002::/16.

Return path selection not optimized.

Security issue if not secured through IPsec.


These mechanisms are presented in this chapter along with more in-depth discussions on deployment recommendations. The recommendations are made only from the perspective of providing IPv6 unicast connectivity. Other selection criteria are presented in the chapters dedicated to other services, such as multicast, virtual private networking (VPN), quality of service (QoS), and security.

Overall, a deployment strategy choice depends on technical considerations (best way to leverage existent IPv4 infrastructure and to circumvent constraints such as Network Address Translation [NAT], for example) as well as cost-effectiveness considerations (for instance, return on the investment of upgrading existent resources or adding new ones). Nevertheless, when possible, it is recommended to begin with native IPv6no IPv6 tunnelsbecause this represents the ultimate goal of the IPv6 integration.


Previous Page
Next Page