Quiet ping is good, busy ping is poor
The line may respond quickly when idle, then games rubber-band, calls freeze or websites hesitate as soon as a large transfer starts.
Loaded latency explained
What Is Bufferbloat?
Bufferbloat is excessive delay caused by traffic queues growing when broadband is busy. It can leave download speed looking healthy while games, calls and ordinary browsing become slow to respond.
Interactive check
Grade Your Loaded-Latency Increase
Enter the quiet idle ping and the highest ping recorded while the connection was busy. The difference is the loaded-latency delta.
Your Result
Latency Delta
— ms
Performance Grade
—
Recommended next step:
Open the Dedicated Bufferbloat Optimisation Guide
Please enter valid values. Loaded ping must be equal to or higher than idle ping.
Recognise the pattern
What Bufferbloat Looks Like During Real Use
The defining pattern is not simply high ping. It is a large increase from quiet latency when downloads, uploads or household traffic fill a queue.
One busy device affects the whole home
Cloud backup, CCTV upload, console updates or a large download can make every interactive application feel delayed while throughput remains high.
Upload often exposes it first
Many UK packages provide much less upstream capacity than downstream capacity, so acknowledgements, voice and gaming packets can queue behind uploads.
Check the UK Average Download and Upload Speed Benchmarks to compare the headroom available on your connection.
Wi-Fi can imitate or amplify it
Weak signal, retries and shared airtime can create a separate wireless queue. A clean Ethernet result with poor room-by-room Wi-Fi points away from the broadband line.
Queue behaviour
Why Oversized Queues Add Delay
Routers and network devices buffer packets to avoid dropping them during short bursts. When traffic arrives faster than the bottleneck can transmit it, an oversized first-in-first-out queue lets bulk data sit ahead of small time-sensitive packets.
Flow queueing changes that behaviour by separating active flows, while Active Queue Management signals congestion before the buffer becomes excessively deep. CAKE and FQ-CoDel combine these ideas to keep delay controlled while the link is busy.
Router or modem queue
The home gateway may buffer more traffic than the broadband uplink can send, particularly when upload capacity is small or variable.
Wi-Fi airtime queue
Frames can wait for airtime or be retransmitted before they even reach the broadband bottleneck.
Mobile or shared-network queue
4G, 5G, cable segments and provider-side links can change capacity, moving the effective bottleneck outside the home.
CPU-limited queue control
A router may offer SQM but lack enough processing headroom to shape a fast package without reducing maximum throughput.
Technical Context: QoS vs. SQMTraditional QoS can prioritise selected traffic without controlling the queue at the true bottleneck. SQM algorithms such as CAKE and FQ-CoDel keep queues shorter, isolate flows and signal congestion earlier by shaping slightly below the stable line rate.
This explainer intentionally stops at the theory. For upstream-specific diagnosis and practical router controls, examine our complete guide to fixing slow upload speeds and setting up router-level smart queue management.
Controlled comparison
Run the Six-Step Bufferbloat Check
Use the same device and change one variable at a time. This integrates the evidence-gathering process with the interpretation instead of separating it into Issue, Validate and Fix stages.
- Measure quiet idle latency. Record baseline ping, jitter and packet loss while downloads, uploads and cloud sync are paused.
- Run a wired control test. Connect directly to the router with Ethernet so Wi-Fi airtime and retries are removed from the result.
- Compare download and upload load. Identify which direction creates the larger latency increase; upload saturation is common on asymmetric packages.
- Repeat with household load paused. A large improvement proves that local traffic volume was filling the queue.
- Compare the problem room over Wi-Fi. Good Ethernet and poor Wi-Fi indicate a local radio problem rather than the provider connection.
- Validate one change at a time. Repeat the same test after each supported queue or bandwidth adjustment so the result proves whether it helped.
Use the result
Choose the Next Step That Matches the Proven Bottleneck
This page defines and identifies queueing behaviour. Detailed device configuration remains in the dedicated troubleshooting guides.
If upload creates the largest spike
Pause background uploads and compare the upstream capacity with the package benchmark. Then use the slow upload troubleshooting guide for upstream saturation and router-control checks.
If Ethernet is poor under load
The queue is likely at the router, modem or upstream bottleneck. Follow the dedicated bufferbloat optimisation guide for supported SQM and shaping steps.
If Ethernet is good but Wi-Fi is poor
Treat Wi-Fi coverage, retries and shared airtime as a separate local problem rather than replacing the broadband package.
If results change by time of day
Repeat identical wired tests. A consistent peak-time pattern can indicate shared provider capacity or a changing external bottleneck.
If gigabit service still grades poorly
High capacity reduces how often the line saturates but does not guarantee good queue management. Large simultaneous transfers can still expose oversized buffers.
If a router change is being considered
Confirm that the replacement supports effective SQM and has enough CPU headroom for the line speed before buying hardware.
Loaded-latency benchmarks
Loaded-Latency Increase Grades
Compare the highest busy ping with the quiet idle baseline. The delta is more useful than either number on its own because it shows how much responsiveness is lost when the connection is loaded.
| Latency Increase | Performance Grade | Likely Real-World Impact |
|---|---|---|
| 0 to 5 ms | Grade A · Excellent | Queues appear tightly controlled. Gaming, calls and browsing should remain responsive while the connection is busy. |
| 6 to 25 ms | Grade B · Manageable | Minor queueing delay. Most everyday use should remain smooth, although fast games may feel slightly less consistent. |
| 26 to 60 ms | Grade C · Degraded | Noticeable delay under load. Games may rubber-band and live calls can develop short freezes or audio disruption. |
| Over 60 ms | Grade F · Serious issue | The queue is adding substantial delay. Interactive applications can become difficult to use whenever large transfers run. |
Detailed answers
Bufferbloat FAQs
Use these answers to understand whether line speed prevents bufferbloat, how to approach gaming problems and what loaded-latency increase should trigger deeper investigation.
Can bufferbloat happen on gigabit full fibre?
Yes, bufferbloat can occur on gigabit full fibre because it is caused by excessive queueing rather than low line speed. Gigabit capacity reduces how often saturation occurs, but large simultaneous transfers can still fill router or upstream queues and increase latency.
How do I fix bufferbloat for gaming?
Confirm the problem over Ethernet, compare quiet and loaded results, then use router-supported queue management or traffic limits one change at a time. Detailed configuration is covered in the dedicated bufferbloat optimisation guide.
What is a good loaded-latency increase?
A rise of 5 ms or less is excellent. An increase of 6 to 25 ms is generally manageable, 26 to 60 ms can disrupt fast gaming or video calls, and anything above 60 ms indicates a serious queueing problem.