Foreword
Multi-Protocol Label Switching (MPLS) began in the mid-1990s with just two modest design objectives. The first was a better integration of ATM with IP, a goal that we hoped could be met by providing a single IP-based control plane that would span both ATM switches and IP routers. The second objective was to augment the IP control plane with some additional functionality, namely traffic engineering using constraint-based routing that was already present in the ATM control plane.
Not long after it started, MPLS usage was extended to applications such as Circuit Cross Connect (CCC), ATM and Frame Relay service over an IP/MPLS infrastructure (draft-martini), BGP/MPLS VPNs (2547 VPNs) and then Virtual Private LAN Services (VPLS). The original constraint-based routing functionality evolved beyond traffic engineering to applications such as fast reroute and Differentiated Services Traffic Engineering (DiffServ-TE).
The idea of a single control plane for both ATM switches and IP routers evolved into Generalized Multi-Protocol Label Switching (GMPLS), which provides a single control plane that could span not only routers and ATM switches but SONET/SDH and optical cross connects as well.
One of the recent MPLS developments deserving of mention here is the use of MPLS in the access network. Expanding MPLS into the access network brings with it scalability challenges. The third edition describes a solution, known as ‘Seamless MPLS’, ...
