At the beginning of this book, we observed that in only a few years MPLS has become a mainstream technology used in a large proportion of service provider networks worldwide. One of the most successful MPLS-based services so far is Layer 3 VPN, now a lucrative revenue earner for many service providers. Also MPLS-based Layer 2 services are becoming available in many regions, either in point-to-point or multipoint form.
One of the reasons why MPLS has developed relatively rapidly is because of the pragmatic way that existing protocols have been adapted and new ones developed to support MPLS. In this book, we saw that an existing protocol, RSVP, was used as the basis for MPLS traffic engineering because of its properties of resource reservation and admission control. Additional properties were added to cater for the requirements of MPLS, such as the ability to distribute labels and specify the path to be followed by the traffic. On the other hand, LDP was developed specifically for MPLS, because no existing protocol had the required properties. An existing protocol, BGP, was adapted to carry the routes (and associated labels) of L3VPN customers, and the same scheme was then carried through to L2VPN and VPLS. A lesson learnt from the development of these VPN services was that no single VPN service type suits all customers so, for example, L3VPN is not inherently ‘better’ than L2VPN or vice versa. Thus in order to address the widest possible range of ...