Scope of the Book
From the discussions in this chapter so far, you know that the design of carrier-class IP/MPLS networks involves reducing both unplanned and planned outages by using a variety of fault-tolerance techniques, including node-level hardware redundancy, control-plane software redun-dancy, MPLS-layer redundant LSPs, OAM mechanisms, and in-service software upgrades. In short, the reliability and availability of an IP/MPLS network encompasses a broad set of functional areas.
The main purpose of this book is to describe IP/MPLS control-plane fault-tolerance mecha-nisms that enable you to reduce downtime and improve network availability (by reducing unplanned IP/MPLS control-plane failures). Specifically, this book intends to cover three aspects of the control plane, as follows:
IP/MPLS forwarding-plane NSF mechanisms that allow a router to continue to forward traffic while its control plane recovers from a failure
IP/MPLS control-plane restart mechanisms that enable IP/MPLS control-plane components to restart and recover state without disrupting the forwarding plane
Use of the previous two mechanisms to reduce downtime in the converged IP/MPLS backbone when using MPLS applications such as traffic engineering (TE), Layer 2 VPNs (L2VPNs), and Layer 3 VPNs (L3VPNs).
In the remainder of this book, it is assumed that the control-plane software executes as a single image containing inseparable nonrestartable components. A detailed discussion of process-level modularity and restartability is beyond the scope of this book.
Although for completeness sake fault-tolerance mechanisms such as MPLS FRR, MPLS OAM, and in-service software upgrades are briefly mentioned in a later chapter, a detailed discussion of these mechanisms is also beyond the scope of this book.