Path computation in Multi Protocol Label Switching (MPLS) and Generalized MPLS (GMPLS) networks has to date been distributed in routers. The emergence of new (G)MPLS applications, such as multicast transport, inter-domain traffic engineering or even inter-layer traffic engineering, leads to complex path computation requirements. In order to cover these requirements the Internet Engineering Task Force (IETF) has introduced a new element in the (G)MPLS architecture, the Path Computation Element (PCE), that computes paths on behalf of routers. Such an externalization of the path computation function allows handling complex computations cases, as for instance the computation of inter-domain paths where a PCE in a domain collaborates with PCEs in other domains. In this paper we firstly provide an overview of the PCE architecture and list its key applications. We then present an experimental PCE implementation  and provide some performance results, before to discuss future applications and challenges.

The introduction of Control Planes (GMPLS, MPLS) with Traffic Engineering (TE) capabilities in co-cs/ps networks has greatly improved the flexibility of such networks;  having the possibility of designing Traffic Engineering policy allows Network Operators to introduces a ‘closed loop’ control of their network resources.  This augmented control of resources can be used to both improve the QoS provided to customer and increase the level of usage of network resources. TE imposes a very careful planning of network, this planning activity is called Network Engineering.  Network Engineering is usually performed by off-line tool, moreover this tool can also be used to simulate the network behaviors in order to perform what if analysis.  This paper illustrates the concept of Network Engineering in a Traffic Engineered network.