Aruba Mobility Expert (Written) common mistakes
Aruba Mobility Expert (Written) · Understand the traps and how to avoid them.
Treating the Mobility Conductor/Mobility Master and the Mobility Controller as a single combined box when reasoning about forwarding paths, so answers about where client traffic is bridged or tunneled are wrong.
Candidates study controller configuration screens and assume all roles execute on one device; the split between management/orchestration plane and data plane is abstract and easy to gloss over.
What to do instead
Draw the control-plane vs data-plane split explicitly for each topology (single controller, conductor + controllers, cluster) and state for every scenario which device terminates the client, which anchors the tunnel, and which holds the configuration master.
Designing a controller cluster without accounting for the standby/backup role assignment and the AP and client failover behavior, then picking an answer that assumes instant seamless failover for all clients.
Cluster concepts (active/standby, LMS/BLMS, fast failover, stateful vs stateless failover) blur together, and candidates assume all failover is stateful.
What to do instead
Memorize which failover keeps client state (stateful) versus requires re-authentication/DHCP (stateless), and for each design question verify the AP's LMS/BLMS assignment and which controller becomes active before selecting an answer.
Choosing a channel and transmit-power plan by picking the highest allowed EIRP or widest channel width instead of applying a reuse plan that minimizes co-channel interference.
Candidates equate wider channels and higher power with better performance and forget that 80/160 MHz reduces the number of non-overlapping channels and raises CCI in dense deployments.
What to do instead
For each RF question, first count available non-overlapping channels in the band, then select width and power that keep co-channel overlap minimal; justify the choice by density and interference, not by raw throughput.
Mixing up 802.1X/EAP roles and where each function lives, e.g., placing the supplicant on the switch or assuming the controller decrypts EAP payloads in all methods.
The supplicant/authenticator/authentication-server triad is taught abstractly, and tunneled vs non-tunneled EAP methods (PEAP/TTLS vs EAP-TLS) differ in where credentials are exposed.
What to do instead
For every auth scenario, label the client as supplicant, the AP/controller as authenticator, and RADIUS as the server, then state whether the method is tunneled and what the controller can see; use that to eliminate wrong options.
Assuming all client VLANs and user roles must be trunked to every controller, or that a role change requires a new subnet, when designing L2/L3 integration.
Candidates conflate role-based policy with VLAN membership and forget that role assignment can be decoupled from VLAN, and that tunneled vs bridged forwarding changes the trunking requirement.
What to do instead
For each integration question, determine the forwarding mode first (tunnel/bridge/split), then derive which VLANs must exist where; remember role derivation can map users to a VLAN or keep them on the same VLAN with different policy.
Diagnosing a client connectivity issue by jumping to RF causes without first checking the ordered verification path (association, authentication, DHCP, ARP, role/VLAN assignment).
RF problems are memorable and over-weighted in study, while the layered troubleshooting sequence is tedious and rarely practiced end-to-end.
What to do instead
Adopt a fixed checklist (client association state, 802.1X/RADIUS result, DHCP offer, ARP/gateway reachability, role and VLAN) and answer troubleshooting questions by locating the first failing layer rather than the most dramatic symptom.
Ignoring log/alert severity and retention implications when choosing a monitoring or maintenance action, such as assuming all events are retained indefinitely or that a syslog level change has no operational cost.
Monitoring features are studied as menu items rather than as operational tradeoffs (storage, noise, escalation), so candidates pick the most comprehensive-sounding option.
What to do instead
For each monitoring question, identify the operational goal (audit, real-time alerting, capacity trending) and match it to the right mechanism and level; reject options that flood alerts or lose required audit records.
Treating advanced services (location, guest/onboarding, AirGroup/mDNS, tunneled node) as independent features without checking their prerequisites such as controller role, licensing, or the forwarding path they depend on.
Advanced topics are learned last and in isolation, so dependencies on base architecture and licensing are missed under time pressure.
What to do instead
For each advanced-service question, state the prerequisite chain (licensing, role, forwarding mode, DNS/mDNS path) before evaluating the answer, and eliminate options that assume a service works without its dependency.
Answering HA and redundancy questions by counting devices rather than verifying that the specific failure scenario (controller loss, link loss, uplink loss) is actually covered by the chosen design.
Candidates equate redundancy with 'more boxes' and do not trace each single point of failure through the design.
What to do instead
For every HA option, walk a specific failure injection (lose controller A, lose uplink on B) and confirm the design survives it with the stated failover type; discard options that leave a single point of failure.
Rushing multi-part scenario questions and answering the first plausible option without reconciling it against the full stated constraints (band, density, security requirement, existing infrastructure).
The exam is long and scenario-heavy; under time pressure candidates pattern-match a keyword to a familiar feature and stop reading constraints.
What to do instead
For scenario items, extract the constraints into a short list before looking at options, then test each candidate answer against every constraint; if two options survive, the one that ignores a stated constraint is the trap.