While choosing the right hardware for a network is critical, more is involved in building a scalable network than simply buying routers with redundant components and a few orders of magnitude more performance than is immediately needed. As important as hardware selection is, it’s only about one-third of the task of building a scalable network. The remaining two-thirds of the task comprise the logical design of the network.
Designing a scalable transport network or adding scalability to an existing design can be accomplished by scaling intradomain routing, by scaling interdomain routing, and by engineering transit traffic through the use of Multiprotocol Label Switching (MPLS). Chapter 12 details the IGP mechanisms that allow the network to grow without putting service or connectivity at risk. Chapter 13 addresses methods of controlling domain-external traffic. This chapter ties the two concepts together to provide a big-picture look at scaling networks.
Any discussion of scaling interdomain routing is, by default, a discussion of BGP scalability mechanisms. This section builds on the Internal BGP (IBGP) configuration basics described in Chapter 13 to show how a simple intelligent route reflection scheme can allow a network to expand or contract to scale as needed.
As explained in Chapter 13, for networks with high availability requirements the authors strongly recommend scaling using route reflection with clusters instead ...