This chapter examined design models for high availability and fast convergence for the hierarchical layers of the Cisco Enterprise Campus Architecture. High availability in the campus minimizes convergence time after link and node failures with appropriate redundancy.
VLANs should not span access switches in the campus for predictable fast convergence. Layer 2 designs use RTSP when STP is required, define primary and secondary root switches, and use the Cisco STP toolkit to harden Layer 2. Trunks and channels are tuned for predictable fast convergence. Aggressive mode UDLD is configured on all fiber links.
Oversubscription and bandwidth are managed to minimize complexity and provide deterministic behavior. Layer 3 designs should load balance traffic over redundant equal-cost links built on triangles, not squares. Routing protocols should peer only on transit links, and summarize at the distribution layer. HSRP and GLBP support fast convergence for end devices.
The Layer 2 to Layer 3 boundary is typically at the distribution layer, but it can be placed at the access layer. Campus network designs should avoid daisy chaining access layer switches, provide appropriate redundancy, and avoid asymmetric flooding.
Infrastructure service considerations such as IP telephony and QoS impact the end-to-end network. The access layer supports device attachment, inline power for devices, and multi-VLAN access ports. End-to-end QoS helps manage oversubscriptions and network speed transitions. Tx-queue starvation is the most common campus congestion issue. Cisco Catalyst Integrated Security features provide security at the network edge.