Network congestion happens when traffic exceeds available bandwidth, causing latency, jitter, and packet drops. Fix it by measuring per-device usage, enabling QoS to prioritise real-time traffic, capping heavy background flows, updating firmware on aging routers, and escalating to your ISP if the bottleneck is upstream rather than local.
Symptoms
- Voice and video calls stutter, freeze, or drop mid-conversation during busy hours.
- Web pages load in bursts, and DNS lookups feel sluggish even on wired clients.
- Cloud backups, Windows Update, or Steam downloads crawl compared to advertised speed.
- Ping and jitter spike above 60 ms to a known-good target such as 1.1.1.1.
- Wi-Fi clients disconnect or roam repeatedly while the SSID still shows full signal.
Common Causes
Oversubscribed uplink
The WAN link is smaller than aggregate demand. A single 4K stream, cloud sync, or OS update can saturate a modest connection during peak hours.
Missing or misconfigured QoS
Without traffic classification, real-time flows like SIP and video conferencing compete equally with bulk downloads and lose.
Broadcast and multicast storms
Loops, misbehaving IoT devices, or unmanaged switches flood segments with broadcast traffic that consumes CPU on every host.
Underpowered or overheating router
Consumer gateways older than five years often cap NAT throughput well below the WAN speed, especially with SPI, VPN, or IDS enabled.
Wi-Fi channel contention
Overlapping 2.4 GHz channels and neighbour APs cut effective airtime. Congestion at the radio layer looks identical to a wired bandwidth issue.
Step-by-Step Fix
- Baseline the connection before you change anything
Run a speed test from a wired client at an idle moment, then again during the reported slowdown. Record ping, jitter, download, upload, and the time of day. Without this baseline you cannot tell whether a change helped or hurt. - Identify the top talkers on your LAN
Log into the router and open the traffic or client statistics view. On Windows, use Resource Monitor's Network tab; on macOS, Activity Monitor's Network tab; on Linux, iftop or nethogs. You are looking for one or two hosts driving most of the throughput. - Check whether the bottleneck is LAN, Wi-Fi, or WAN
Copy a large file between two wired clients to test LAN throughput. Repeat over Wi-Fi to isolate the radio. Then run a speed test against your ISP's own server. Each test narrows the fault domain and prevents you from tuning the wrong layer. - Enable QoS and classify real-time traffic
In the router's QoS or SQM section, set your measured upload and download speeds as the caps, then mark voice, video conferencing, and interactive gaming as highest priority. Bulk categories such as file sync and OS updates go to the lowest class. Reboot and retest. - Reduce Wi-Fi contention
Use a Wi-Fi analyser to find the least crowded 2.4 GHz channel (1, 6, or 11) and an 80 MHz block on 5 GHz. Disable legacy 802.11b rates if your clients support n or later. Move bandwidth-hungry devices to 5 GHz or wire them. - Cap or schedule bulk workloads
Throttle cloud backup clients, Steam, and Windows Update to run overnight or at a fixed rate. Most agents expose bandwidth limits in their settings. On the router, use per-device rate limits for guest devices and IoT. - Update firmware and consider capacity
Apply the latest stable firmware from the vendor's support site. If the router's CPU stays above 80 percent under load, or your WAN plan is saturated even after tuning, the fix is hardware or a larger plan, not more configuration. - Escalate to the ISP with evidence
If the WAN link tests slow directly from a modem-connected laptop, open a ticket with timestamps, speed test URLs, and traceroutes. Ask specifically about upstream congestion, CGNAT saturation, or a noisy last-mile line.
Symptom to likely cause to first action
| Symptom | Likely cause | First action |
|---|---|---|
| Calls drop only during backups | Uplink saturation | Cap backup upload rate |
| Slow Wi-Fi, fast Ethernet | Radio contention | Change channel, move to 5 GHz |
| Everything slow at 8pm daily | ISP peak-hour congestion | Test from modem, contact ISP |
| Router hot to touch, high latency | Overloaded gateway CPU | Update firmware or replace unit |
| One PC saturates the LAN | Runaway sync or malware | Inspect processes, isolate host |
| High ping, low throughput on all clients | Bufferbloat | Enable SQM or fq_codel QoS |
Prevention
- Review router traffic dashboards monthly to spot new bandwidth-heavy devices early.
- Keep firmware current and replace consumer gateways roughly every four to five years.
- Segment IoT and guest traffic onto a separate VLAN or SSID with its own bandwidth cap.
- Schedule backups, patch cycles, and large syncs outside working hours.
FAQ
How do I tell network congestion apart from a slow ISP plan?
Run a speed test directly from a laptop plugged into the modem, bypassing the router and Wi-Fi. If results match your plan, the bottleneck is inside your network. If they fall short at multiple times of day, the plan or the ISP line is the real limit and no amount of QoS will fix it.
Does adding a mesh system reduce congestion?
Mesh helps with Wi-Fi coverage, not with WAN bandwidth. If your uplink is already saturated, mesh will not add capacity. It can reduce airtime contention in large homes or offices by moving clients closer to a node, which improves effective throughput on the wireless leg only.
Is QoS still useful on gigabit connections?
Yes, especially for upload-heavy real-time traffic. Even on gigabit, a single cloud upload or backup can fill the upstream and add hundreds of milliseconds of bufferbloat. Modern SQM disciplines such as CAKE or fq_codel keep latency low under load, which matters more than raw speed for calls and gaming.
Book a network assessment with our IT team to diagnose and remediate persistent congestion.