PoE Budget Blocks a Wi-Fi 7 Refresh
Bayline Insurance is refreshing 26 ceiling access points in one closet of a 2016 office building. The existing switch is a 48-port 802.3at unit with a 370 W PoE budget; the current APs draw 12.9 W each (26 × 12.9 = 335 W), leaving 35 W of headroom. The new tri-band Wi-Fi 7 APs are specified for 802.3bt Type 3 for full features, and the data sheet states that under 802.3at they present the 6 GHz radio at 2x2 instead of 4x4 and disable the 10G port. Horizontal cabling is Cat5e, the switch uplink is 1 GbE, and the closet UPS is rated 1,500 VA with 22 minutes of runtime at present load. Capital for the AP refresh is approved; switch and cabling replacement need a separate business case due in six weeks. Facilities require any new cable pathway to be fire-stopped, which adds about three weeks. About 35% of associations today are from 6 GHz-capable clients.
1. Quantify the power problem and state what the refresh actually requires from the switching layer.
Read the worked answer
At 802.3bt Type 3 the switch must budget roughly 60 W at the PSE per port, so 26 APs need about 1,560 W of PoE budget; a 370 W switch supports at best 14 APs at 802.3at and cannot power the full refresh at all. The new switch therefore needs a PoE budget of roughly 1,500–2,000 W to leave 20–25% headroom, which is a different class of unit and a different electrical problem: a 2 kW load on a 120 V circuit is close to a 20 A breaker's continuous rating, so the electrical panel and the PDU become part of the scope, and the UPS must be resized rather than assumed adequate. Note also that the 10G port is disabled under 802.3at anyway, and Cat5e supports 2.5G and 5GBASE-T to 100 m but not reliable 10GBASE-T, so full-speed uplinks would require the fire-stopped pathway and Cat6A. The decision is not 'buy bigger switches'; it is a power, electrical and cabling project that must be authorised as one.
2. Given the six-week business case delay, what interim option would you recommend, and how would you justify accepting reduced 6 GHz capability if you take it?
Read the worked answer
Do not half-refresh a floor. Powering 14 new APs and leaving 12 old ones on the same switch produces a mixed coverage and capability footprint, and the new APs will be the ones dropping their 6 GHz radio to 2x2 while colliding with a neighbour that is not in the same channel plan. Two defensible interim paths exist. First, a fixed-function 802.3bt switch for a single pilot zone, so the new hardware is validated in production while the business case is written. Second, defer the refresh and write the business case around measured data. If the reduced mode must be accepted, justify it with numbers: 35% of associations are 6E-capable, and a 2x2 6 GHz radio costs roughly 3 dB of receive gain and half the peak PHY rate relative to 4x4, which is tolerable in an office with low client density but not in the conference floor. Put that trade-off in writing with a review date rather than letting it become permanent.
3. What would you verify after cutover, and which monitoring would you put in place for the first month?
Read the worked answer
Verify the physical layer before celebrating the APs: measure actual per-port power draw against the data sheet rather than trusting nameplate sums, confirm PoE budget headroom stays above 20% with all APs at peak, check closet temperature and airflow because 2 kW of PoE becomes heat in the same room, and recalculate UPS runtime at the new load instead of assuming 22 minutes still holds. On the network, watch port errors, discards and retransmits on the uplinks, and confirm the uplink is not the new bottleneck once 6 GHz clients appear. For the first month, trend per-radio client counts, retries, channel utilization, 6 GHz association share, and AP reboots or PoE faults, since an under-budgeted switch typically shows up as APs rebooting or ports dropping under load rather than as a clean failure. Compare against a pre-change baseline of client experience, not just AP health.