Bufferbloat in gaming: the test that proves streaming causes your lag

Part of our guide on high packet loss low ping

A quick ping test while streaming shows if queued delay in your router is the culprit and points to fixes.

What bufferbloat is and why it causes lag when streaming

Bufferbloat is delay created when outgoing packets wait in a router's transmit queue. The queue sits between your LAN and the ISP link. When uploads congest that link, game packets can wait behind large amounts of data and get delayed.

Queued packets cause higher latency and jitter. That shows as sudden ping spikes, shot delays, rubber-banding, and stuttering in online games when someone streams or uploads heavily.

How big the delay gets depends on three things: the router's buffer size, the upload bandwidth, and how the router manages queues (firmware). A large buffer filled at a given link rate takes longer to drain, so delay grows roughly as buffer size divided by link rate. For example, a 2 MB buffer on a 2 Mbps link can introduce around eight seconds of queuing if fully filled; that number changes with link speed and real buffer behavior. Use active tests outlined below to measure realistic delay for your setup.

Some routers include Active Queue Management (AQM) algorithms like fq_codel that keep queues short under load. Other routers use big buffers or poorly tuned queuing, which lets delay climb into the tens or hundreds of milliseconds. Which case applies to you depends on the specific router model and firmware. See the router-identification section for how to check this in your hardware. We go through outgoing packet loss fixes step by step elsewhere on the site.

Sources and tests. Researchers and hobbyists document bufferbloat measurements using tools such as Flent and the DSLReports bufferbloat test. OpenWrt and network engineering notes explain fq_codel and SQM behavior. These sources show variable but repeatable reductions in latency when AQM is enabled.

How to test your network for bufferbloat impact during gaming

You need a controlled comparison. Measure latency to your game server with the network idle, then run the same test while the stream or upload is running. A consistent jump in latency under load shows queued delay.

Tools to use. Run Flent (for scripted, repeatable tests) or the DSLReports bufferbloat test. If you want a simple method, run continuous ping to your game server or to a reliable IP like your ISP's router while you start a stream that uses your usual upload bitrate. We go through how to change mtu on router step by step elsewhere on the site.

What to look for. Compare baseline latency and loaded latency. Watch for spike size and jitter. If latency climbs by tens to hundreds of milliseconds and stays high while upload is saturated, bufferbloat is likely the cause.

Testing upload levels. Repeat the test at different upload rates: 30%, 60%, 90% of your link. Note the point where latency rises sharply. That point gives you an estimate of the safe upload ceiling for your setup.

Interpretation guidance. Exact numbers vary. Some setups show small increases until near full saturation. Others show big spikes at moderate load because of large buffers or missing AQM. These tests let you quantify the problem for your own hardware and ISP. The other half of this decision is understanding ping vs traceroute.

1. Run a baseline ping to your game server with no uploads.
2. Start a controlled upload or stream at a fixed bitrate.
3. Run the ping test while upload runs and note latency and jitter.
4. Repeat at other upload rates to find the point where latency becomes unacceptable.

What changes to make if bufferbloat is confirmed

Limit upload usage while gaming. Lower your stream's bitrate or pause large uploads. That prevents queues from forming and is the simplest fix.

Enable QoS or SQM on your router. SQM implementations using fq_codel actively manage queues to keep latency low. If your router UI shows 'Smart Queue Management', 'SQM', 'AQM', or 'fq_codel', enable it and apply a conservative setting for upload rate (a small headroom below your measured max helps).

If your router lacks these features, install firmware such as OpenWrt that supports SQM and fq_codel, if the device is compatible. It helps to understand openwrt qos settings to reduce game lag and packet loss before going further.

If firmware changes are not an option, limit upload in the streaming software. Set a bitrate that stays below the upload threshold found in your tests.

Always re-test after changes. Use the same ping or Flent tests to confirm that latency under load falls to acceptable levels.

• Upload limit: set streaming bitrate so upload stays under the threshold you identified (often under 60–80% of max upload).
• Router SQM: enable fq_codel or similar AQM and set upload rate slightly below actual link speed for best results.
• Firmware: consider OpenWrt or vendor firmware with SQM if your router supports it.
• Testing: re-run tests after each change to verify improvement.

How to identify routers and firmware that support AQM or fq_codel

Check the router's web interface for settings named 'AQM', 'Smart Queue Management', 'SQM', 'fq_codel', or 'Advanced QoS'. Those names indicate built-in queue management. If that sounds like your situation, read up on when 5ghz wifi lowers latency next.

Consult the router's manual or spec sheet for 'Smart Queue Management' or queueing algorithm names. If the spec sheet does not list AQM, the device may not support modern queue control.

Search the OpenWrt hardware compatibility list. Many consumer routers run OpenWrt and gain SQM support via its package 'sqm-scripts', which uses fq_codel. The OpenWrt wiki describes supported devices and setup steps.

Check community resources and vendor pages for examples. For instance, many recent consumer routers from major brands have firmware updates or menus that expose SQM or 'Gaming QoS'. If a model's community support shows OpenWrt builds, you can add SQM there.

How to confirm in practice. Enable the feature and run the same bufferbloat test. If latency under load drops significantly, the AQM implementation works for your case.

Practical examples, assumptions, and variability

Avoid trusting a single number. Published experiments show different results because buffer sizes, link rates, and real-world traffic differ across homes and ISPs. Use tests to see how your setup behaves.

Illustrative calculation with clear assumptions. Suppose a router's transmit buffer holds 2 megabytes of full-packets worth of data and your upload link is 2 megabits per second. If the buffer fills, the queuing delay would be roughly buffer size (in bits) divided by link rate. Two megabytes equals 16 megabits, so 16 megabits divided by 2 megabits per second gives about eight seconds of potential queuing delay. This is a math illustration, not a typical claim about all routers. Many routers use smaller buffers or AQM, which drastically reduces queuing time.

Why this matters. The same buffer size on a faster link yields much less delay. A 2 MB buffer on a 100 Mbps upload link would add far less latency. That explains why bufferbloat varies strongly with both buffer size and upload speed.

Sources. For measurements and protocol details, see the DSLReports bufferbloat tests and Flent examples for bufferbloat measurement. OpenWrt documentation covers SQM and fq_codel implementation and tuning. These resources provide details and community-measured results.

Advice when troubleshooting in your home network

Start with tests. Measure before you change settings. That gives a baseline to compare after fixes.

Work incrementally. First lower the stream bitrate, then enable SQM or QoS, then consider firmware or hardware changes.

Keep expectations realistic. Some ISP setups or modem/router combinations limit what you can change. In those cases, limiting upload from the streamer is often the most reliable fix.

Document your setup. Write down router model, firmware, and the test results so you can search for device-specific guidance or ask for help on forums with accurate data.

Questions people still ask

I have run repeated bufferbloat tests on home connections and used OpenWrt SQM with fq_codel to reduce latency under upload load.