The 4 p.m. Slowdown at Beacon Tutorial Centre

The 11:00 am test measures the line when almost nobody is using it. It shows that one device can reach 92 Mbps at that moment. It says nothing about 4:00 pm, when 40 laptops, staff phones and students on guest Wi-Fi compete for the same access points and the same 100 Mbps uplink. A single speed test also collapses several different quantities into one number.

Measurements to take during the slowdown:

• Throughput per SSID and per access point, from a wired PC and from a wireless client. This separates a saturated WAN line from saturated Wi-Fi.
• Latency, jitter and packet loss to the LMS and to the video service. Freezing video is usually delay or loss, not raw bandwidth.
• DNS lookup time. A slow or overloaded DNS server makes every page feel slow even on a fast line, and explains the printer's web page failing.
• Wi-Fi channel utilisation, signal strength and client counts per AP. Twelve APs across three floors can interfere with each other and with neighbouring networks.
• DHCP pool usage, since a guest network with a short lease pool can run out of addresses at peak time.

Without monitoring, the 4:00 pm event cannot be attributed to a cause, so the upgrade is being bought blind.

Bandwidth is the maximum data per second the line can carry. Freezing video is usually caused by latency, jitter or packet loss, and by contention for a shared medium - none of which a bigger pipe fixes.

• Contention: if students stream video on guest Wi-Fi, they can fill whatever capacity exists. A 1000 Mbps line with no prioritisation still lets a few heavy users create queues that staff video traffic sits behind.
• Wi-Fi airtime: each access point serves a limited number of clients, and all clients on one channel share airtime. Forty laptops plus phones on twelve APs in a dense building can hit that ceiling regardless of the WAN link.
• Latency and jitter: a video call needs steady, timely delivery. If packets queue behind bulk downloads, the call freezes even though average throughput looks fine.
• Local network limits: if the router's LAN ports or the access point uplinks are 100 Mbps, wireless clients see no improvement at all from a 1000 Mbps line.
• Beyond the router: the LMS or video provider may itself be slow at peak times, and slow DNS adds delay to every new connection.

So the upgrade only helps if the WAN link is proven to be the bottleneck, and even then only with traffic management in place.

Priority order:

1. Measure first. Run the 4:00 pm tests above for one week. If the 100 Mbps line is not saturated, the upgrade is not the answer.
2. Separate staff and guest traffic onto different SSIDs and VLANs, and put staff desktops on wired connections where possible. Guest users then cannot crowd out teaching traffic.
3. Apply QoS so video conferencing and the LMS get priority, and cap bandwidth per guest client.
4. Tune the Wi-Fi: move APs off congested channels, reduce transmit power to shrink overlapping cells, and check that each AP has a wired uplink of at least 1 Gbps.
5. Add a local DNS cache or use a faster resolver, which addresses the slow page loads and the printer page.
6. Add simple monitoring so the next complaint can be attributed to a cause rather than guessed at.

Buy the 1000 Mbps upgrade only if the tests show the WAN link saturated at 4:00 pm and the switches and AP uplinks can carry the higher rate. Even then, keep the QoS and guest separation - otherwise the extra capacity will be consumed the same way.