How the test measures bufferbloat
The test measures ping three ways:
- Idle: 21 small requests before any heavy traffic starts.
- While downloading: a small request every 400 ms or so while large downloads run.
- While uploading: the same, during the upload test.
The large number in your result is the higher of the two loaded pings. The difference between that and your idle ping is the delay that builds up in queues when your connection is full. You can see both loaded values under “Test details”.
Reading the increase
| Increase over idle ping | What it means in practice |
|---|---|
| Under 30 ms | Little or no bufferbloat. Calls and games hold up when others are downloading. |
| 30–100 ms | Moderate. You may notice lag and slower page loads while someone is downloading or uploading. |
| Over 100 ms | Significant. Games and calls will suffer whenever the line is busy. |
These bands are our general guidance; there is no official standard. Upload bufferbloat is often worse than download on cable and DSL, because upload capacity is smaller and fills up faster.
How to fix bufferbloat
The proven fix is smart queue management (SQM) on your router. SQM algorithms such as fq_codel (described in RFC 8290) and CAKE keep queues short and share them fairly, so a small game or call packet does not wait behind a big download.
- Look in your router settings for SQM, Smart Queue, fq_codel, CAKE, or a QoS option that asks for your line speed.
- Set the limits slightly below your measured speeds, often 5–10% lower, so the queue forms in the router where it can be managed.
- Retest. The loaded ping should be much closer to idle.
If your router has no such option, a router that supports SQM is often a better upgrade than a faster plan. More detail is in what bufferbloat is and how to reduce ping.