An analytical framework for the performance of different protection switching restoration mechanisms with different protection policies in Multi-Protocol Label Switching (MPLS) networks is developed, where the protection policy may be 1 + 1, 1:1, n:l or n:m. As a general rule, it aims at the MPLS networks with the protection paths created in advance of fault occurrence without resources reserved, and it reroutes the protected traffic in the event of link or router failures while trying to maintain quality of service (QoS) requirements. Based on the proposed protocol for processing faults, an analytical model and an evaluating model are created, and the benefits and penalties of different mechanisms with different protection policies are evaluated. In particular, an improvement on the proposed protocol is presented to overcome the drawbacks of n:m protection policy in which m > 1 and n> 1.