High Latency on WAN Link – Fix It in 3 Steps

The Quick Fix (30 Seconds)

Before you go changing router settings, check the damn cables. I've seen this a hundred times. A loose Ethernet cable or a bad splitter on a coax line will spike your latency by 50-100ms. Unplug and replug every cable between your modem and router. If you're on cable internet, screw the coax connector back on tight—hand-tight, don't use pliers.

Also, reboot everything. Pull the power on your modem and router for 30 seconds. Plug the modem in first, wait 2 minutes till it syncs, then plug in the router. This clears bufferbloat in your modem's memory. Test with a ping to 8.8.8.8 -t on Windows or ping -c 10 8.8.8.8 on Mac/Linux. If latency drops below 20ms, you're done.

Real-world trigger: This happens all the time when someone moves furniture and kicks a cable loose. Or after a thunderstorm—moisture in outdoor coax crimps your signal. Always check physical connections first.

The Moderate Fix (5 Minutes)

If the quick fix didn't work, you've got bufferbloat. That's when your router's queue gets too deep and causes latency spikes under load. The fix: enable Smart Queue Management (SQM) or QoS on your router.

For most home routers (Asus, TP-Link, Netgear), log in to 192.168.1.1 or 192.168.0.1. Find the QoS section. Enable it and set your WAN upload and download speeds to 90% of your plan's rated speed. So if you pay for 100Mbps down, set it to 90Mbps. This prevents the buffer from filling up.

If you're on OpenWrt or pfSense, install the Cake or fq_codel queue discipline. Command for OpenWrt:

opkg install luci-app-sqm
# Then configure via LuCI: Services > SQM QoS

pfSense users: go to System > Advanced > Networking, enable "Traffic Shaper" with the fq_codel scheduler. Set bandwidth to 90% of your measured speed. Run a speed test while pinging Google—if your ping stays under 30ms, you fixed it.

Skip the "gaming mode" crap on consumer routers. It rarely helps. fq_codel or Cake is what actually works.

The Advanced Fix (15+ Minutes)

If you still see latency spikes of 100ms+ during peak hours, it's probably your ISP's problem. But before you call them, eliminate your gear.

Step 1: Isolate the modem. Plug a laptop directly into the modem (not the router). Run a continuous ping to 1.1.1.1. If latency is stable, your router is the bottleneck. If it still spikes, it's the modem or ISP line.

Step 2: Check for noise. Cable modems are sensitive. Log into your modem's status page (usually 192.168.100.1). Look at the Downstream Power Level—should be -7 to +7 dBmV. If it's outside that range, call your ISP and say "my modem reports power levels out of spec, send a tech." Upstream SNR should be above 30 dB. If it's under 25, you've got ingress noise. That means a bad splitter, damaged cable, or neighbor's crappy wiring.

Step 3: Change DNS. This is a long shot, but slow DNS servers can add 20-30ms to every request. Set your router to use Cloudflare (1.1.1.1) or Google (8.8.8.8). I prefer Cloudflare—they log less.

Step 4: Check for saturation. If your WAN link is saturated (99% usage), latency goes to hell. Run a bandwidth monitor like ntopng or just watch Task Manager's network tab. If upload or download is pegged, throttle heavy apps. On Windows, turn off OneDrive syncing during work hours. On Mac, limit iCloud backup times.

Step 5: Replace the router. If you've done everything above and still get 50ms+ spikes on a fiber line, your router's CPU is dying. I've replaced dozens of old Netgear R7000s that couldn't handle modern cable speeds. Buy a router with a fast CPU—Ubiquiti EdgeRouter 4 or a PC running pfSense. Spend $100-200 and never have this issue again.

If none of this works, it's your ISP's upstream congestion. Call them with a log of ping spikes showing times. Say "I've tested with a wired connection, rebooted your modem, and checked my hardware—you need to check your node." They'll run a line test. If they blame your router, plug directly into the modem and repeat the test. If latency is still high, demand a tech visit.