Fast failure recovery is an important issue in modern communication networks. Critical services, such as distant medical services, stock exchange systems, online gaming, are very sensitive to network disruptions. Conventional IP networks rely on lower layer mechanisms (e.g., SONET automatic protection switching and/or MPLS fast reroute) to achieve fast failure recovery. Recently, a new approach called IP fast reroute (IPFRR) has been introduced to enable fast failure recovery in the IP layer. In this talk, I will introduce the basic principle of IPFRR. After that, I will explain our research on IPFRR, focusing on single-link and shared-risk link group (SRLG) failure scenarios. Our work on SRLG received the 2010 ICCCN best paper award in Zurich, Switzerland.