Implement Roaming, Redundancy, and High Availability
Roaming, redundancy, and high availability
Roaming is a client decision, but the infrastructure must make it fast and clean. Layer 2 roaming keeps the client on the same VLAN and subnet; Layer 3 roaming preserves the client IP across subnets using home-agent/foreign-agent tunneling in a mobility domain. Fast transition (802.11r), OKC, and PMK caching reduce reauthentication time; they are most useful when the client crosses controllers or mobility domains. Design overlapping coverage so clients roam before signal degrades-typically -67 dBm at cell edges with 15-20% overlap. Band steering and load balancing can help, but aggressive settings cause sticky clients or failed associations.
For redundancy, use controller clustering and HA. APs terminate on a cluster and are served by a primary and backup controller; stateful synchronization lets a standby take over. VRRP provides a virtual IP for management and services. Ensure both controllers have the same software version, licenses, VLANs, and regulatory settings, and that the mobility domain matches. For L3 roaming, the home agent must remain reachable; if it fails, clients may lose their anchor.
Traps: assuming 802.11r is required for all roaming; forgetting that L3 roaming requires a mobility domain and anchor controller; and building HA with mismatched licenses or VLANs.
Worked example
Two controllers in a cluster serve 100 APs. Each AP has an active LMS and a standby controller. If the active controller fails, the AP detects the heartbeat loss and fails over; clients reauthenticate or use cached keys. Validate by failing one controller in a maintenance window and checking AP and client recovery.