The 9 A.M. Video Freeze
A school has a 100 Mbps internet uplink. At 9:00 a.m., 30 classrooms join live video lessons at about 2.5 Mbps each, and a cloud backup job begins transferring 60 Mbps. Teachers report video freezing and a jump in round-trip latency from 15 ms to 300 ms. A ping test at noon, with no video and no backup, shows 15 ms and no packet loss. The LAN has a 48-port switch and a router that places all outbound packets in one FIFO buffer. The IT director proposes buying a second 48-port switch to add capacity.
1. Compute the uplink utilization during the backup window under the stated assumptions. Does the link need to be 100% utilized for the video calls to fail? Explain.
Read the worked answer
Video demand is 30 x 2.5 = 75 Mbps. Adding the 60 Mbps backup gives 135 Mbps offered to a 100 Mbps uplink, so utilization is 135%, which is oversubscribed. The link cannot carry all offered traffic, so packets queue or are dropped.
Even below 100% average utilization, video can fail because traffic is bursty. When bursts arrive faster than the uplink can drain them, the FIFO buffer fills, and every packet in the queue waits. Queueing delay can reach hundreds of milliseconds before average utilization reaches 100%. That matches the reported 300 ms latency with no packet loss at noon. The noon test only proves the path is healthy when idle; it does not test the congested condition.
2. The IT director proposes replacing the 24-port switch with a 48-port switch. Evaluate this as a fix.
Read the worked answer
The proposal does not address the bottleneck. The switch provides LAN ports; the constraint is the 100 Mbps internet uplink and the single FIFO queue on the router. More ports may even worsen congestion by making it easy to attach more demand. The reported symptom is latency under load, which is a queueing problem at the uplink, not a port-density problem.
A useful fix must reduce offered load, prioritize latency-sensitive video, or improve queue management. Examples include rate-limiting or rescheduling the backup, enabling QoS so video is served before bulk transfers, and using fair queueing or AQM on the router. The director should first measure per-class throughput and queue delay, then choose a fix that targets the uplink.
3. Propose a two-part mitigation that does not require buying more bandwidth, and predict one side effect.
Read the worked answer
Part 1: cap the backup job at, say, 20 Mbps and move it outside the 8:00-10:00 lesson window. That reduces offered load from 135 Mbps to 95 Mbps at peak and removes the sustained overload. Part 2: enable QoS on the router, marking video conferencing as high priority and bulk backup as low priority, with fair queueing or AQM so a burst does not fill a single FIFO buffer.
Side effect: the backup will take longer and may miss its completion window, so the district must monitor it or allow it to resume. QoS can also starve bulk traffic if misconfigured, so the low-priority class needs a guaranteed minimum share. Neither change adds bandwidth, but together they control queueing and protect interactive video.