Summary

  • RFC 5416 defines a CAPWAP binding for the IEEE 802.11-2007 baseline, not a universal contract for later Wi-Fi features.
  • An advertised radio, Split-MAC mode, roaming message or key-refresh exchange is protocol evidence, not proof of feature support or client continuity.
  • Production decisions need the binding identity joined to accepted configuration, capability evidence and current user-facing telemetry.

A binding is a boundary, not a product promise

The RFC centralizes pieces of 802.11 control between a Wireless Termination Point and an Access Controller. It describes station configuration, radio information, QoS, BSSID-to-WLAN mapping and control messages. That specificity is valuable precisely because it is finite. The binding supports the 802.11-2007 baseline. It can describe an 802.11n radio, but it does not define how to configure or use the additional 802.11n features. The text also leaves 802.11v and 802.11r outside this binding. A capability field therefore identifies a protocol-era relationship; it is not a purchase-order for every later feature.

Advertise is not apply

An operator may see a radio advertised, a WLAN mapped to a BSSID and a station configuration acknowledged. Those records answer “what did this binding report?” They do not answer whether the intended policy was accepted by the hardware, whether the policy is still current, or whether an authorization decision survived a controller change. The control record must be joined to a configuration version, device capability evidence and an approval decision. Otherwise an old, well-formed response can be mistaken for current authorization.

Roaming and keys are separate evidence

RFC 5416 gives roaming and group-key refresh a place in the control exchange. That does not make a roaming notification equal to a successful handoff for every client. A key-refresh message can be correctly formatted while a client rejects the result, loses its session or experiences a service interruption. Measure association continuity, reauthentication, packet delivery and user-visible latency beside the binding messages. The distinction is operational: a protocol event is a checkpoint in an evidence chain, not the outcome at the edge.

Split and Local MAC do not settle the data path

Split-MAC and Local-MAC modes divide work between the access point and controller. They help an operator reason about where control and forwarding decisions belong, but the selected mode cannot certify the path a client actually used. A stable controller record can coexist with a radio fault, a congested backhaul or a failed roam. The mode is an architectural fact; service health still needs independent data-plane measurements.