You have a 200 Mbps connection. Your speed test confirms it. Yet the moment someone in your house starts downloading a game update or uploading a large file, your video call freezes, your game starts rubber-banding, and every webpage takes forever to load. The connection is not slow — the speed numbers prove it. But something is clearly wrong.

This problem has a name, and it is called bufferbloat. It is one of the most common and least understood causes of internet lag in homes and small offices. The frustrating part is that most people never identify it because their speed test results look perfectly normal. This article breaks down exactly what bufferbloat is, how to confirm it is your issue, and how to fix it — often without buying any new hardware.

What Bufferbloat Actually Is — Why Your Speed Test Looks Fine But Everything Lags

Bufferbloat is not a bandwidth problem. It is a queue management problem inside your router. Understanding the difference is critical because it explains why adding more speed to your plan will not solve it.

What Happens Inside Your Router When Someone Starts a Large Download

Every router has a memory buffer — a temporary holding area for network packets waiting to be sent or received. When your connection is idle or lightly used, packets move through this buffer almost instantly. Latency stays low, and everything feels responsive.

The problem begins when a device on your network saturates the connection — a large Steam download, a cloud backup uploading gigabytes, or a 4K stream consuming most of your available bandwidth. When that happens, the router’s buffer fills up with packets from the heavy transfer. Every other device’s packets — your Zoom audio, your game inputs, your DNS lookups — get stuck at the back of that queue. They are not dropped. They are delayed, sometimes by hundreds of milliseconds.

That delay is what you experience as lag. Your connection is still fast in raw throughput. But the real-time traffic that needs to arrive in under 50 milliseconds is now waiting 300 to 800 milliseconds behind a wall of bulk download packets.

Why Bufferbloat Never Shows Up in a Normal Speed Test

A standard speed test only measures throughput — how many megabits your connection can push per second. It runs on an idle connection, pushes maximum data, and reports a number. It does not measure what happens to latency while the connection is under load.

This is why your internet can feel painfully slow even when your speed test shows full speed. Bufferbloat is a latency-under-load problem. It only shows up when the connection is actively being used by multiple things at once. A regular speed test will never catch it because it tests throughput in isolation.

The Exact Symptoms That Confirm Bufferbloat Is Your Problem

If you recognize more than one of these, bufferbloat is almost certainly the cause:

• Video calls freeze or pixelate the moment a download starts on another device.
• Online gaming becomes unplayable when someone else is streaming or uploading.
• Webpages load slowly during active file transfers, even though your plan speed is high.
• Voice calls on apps like WhatsApp or Discord get choppy specifically during heavy network use.
• Everything returns to normal the instant the download or upload finishes.

The key pattern is that lag is tied directly to network load. When nothing heavy is running, everything is fine. The moment bandwidth gets saturated, latency spikes and real-time applications break. That pattern is the signature of bufferbloat — not a slow connection, not a Wi-Fi issue, and not a problem your ISP will find on their end.

How to Test If You Have Bufferbloat Right Now

Recognizing the symptoms is a strong indicator, but you need a measurable result to confirm bufferbloat and to compare before and after applying any fix. A single tool can give you a clear, graded answer in under 30 seconds.

How to Use the Waveform Bufferbloat Test — Step by Step

The most reliable browser-based test for this is the Waveform Bufferbloat Test. It works differently from a standard speed test because it measures latency while simultaneously pushing your connection to full load — exactly the condition where bufferbloat reveals itself.

To get an accurate result, follow these steps:

1. Connect your computer to your router using an Ethernet cable. Testing over Wi-Fi can introduce variable latency that skews the result.
2. Stop all active downloads, streams, uploads, and cloud syncs on every device in your network. The test needs to be the only thing saturating the connection.
3. Open the Waveform test page in your browser and click the start button.
4. Let the test complete fully. It runs a download phase and an upload phase while continuously measuring latency throughout.

The test takes roughly 20 to 30 seconds. When it finishes, you get a letter grade along with detailed latency numbers for both download and upload directions.

How to Read Your Bufferbloat Grade — A to F Explained

The test assigns a grade based on how much your latency increases under load:

• Grade A — Latency stays low even at full throughput. Bufferbloat is not a problem on your connection.
• Grade B — Minor latency increase under load. Most users will not notice issues.
• Grade C — Noticeable latency spikes. Video calls and gaming may occasionally stutter during heavy transfers.
• Grade D — Significant bufferbloat. Real-time applications will consistently degrade when the connection is loaded.
• Grade F — Severe bufferbloat. Latency spikes of 300 ms or more under load. Video calls will freeze, gaming will be unplayable, and even basic browsing will feel sluggish during any large transfer.

What Your Latency Numbers Mean During the Test

Below the grade, the test shows latency values in milliseconds. The number that matters most is the latency increase during the loaded portion of the test — not your idle baseline.

An idle latency of 10 to 20 ms is typical. If that number jumps to 150 ms, 400 ms, or even over 1000 ms during the download or upload phase, your router is holding packets in an oversized buffer instead of managing them. That spike is the direct measurement of bufferbloat. It is also the reason video calls lag on connections that should be fast enough. The raw bandwidth is there, but the jitter and delay introduced by the bloated buffer make real-time communication impossible.

What Grade D or F Actually Means for Your Daily Internet Use

A Grade D or F is not just a theoretical concern. It translates directly into the problems described earlier — choppy Zoom meetings, unplayable online games, and slow-loading pages during any active transfer. If your household has more than one person online at the same time, a D or F grade means someone is always experiencing degraded performance.

The important takeaway is that this problem exists at the router level. Your ISP delivers the bandwidth. Your router mismanages the queue. That distinction matters because it means the fix is almost always within your control.

Fix 1 — Enable QoS on Your Existing Router

If your bufferbloat test came back with a Grade C, D, or F, the first fix to try is enabling QoS on your current router. Most modern routers include some form of QoS, and when configured correctly, it can dramatically reduce latency under load without requiring any new hardware.

What QoS Does and How It Reduces Bufferbloat

QoS stands for Quality of Service. At its core, it controls how your router prioritizes and schedules network packets. Without QoS, your router treats every packet equally — a massive file download and a tiny Zoom audio packet sit in the same queue, first come first served. The bulk transfer floods the buffer, and the real-time traffic waits.

When QoS is enabled and properly configured, the router actively manages the queue. It limits the total throughput slightly below your connection’s maximum capacity, which prevents the buffer from filling up completely. This keeps a small amount of headroom available so that time-sensitive packets — voice, video, game inputs — can pass through without sitting behind hundreds of bulk data packets.

The result is that your download still runs fast, but your latency stays controlled. The router stops blindly stuffing packets into an oversized buffer and starts making intelligent scheduling decisions.

How to Find QoS Settings on Your Router

QoS settings are located in your router’s admin panel. Access it by typing your router’s gateway IP into a browser — usually 192.168.1.1 or 192.168.0.1. Log in with your admin credentials.

Once inside, look for QoS under one of these common menu locations depending on your router brand:

• ASUS — Adaptive QoS or Traffic Manager in the left sidebar.
• TP-Link — Advanced → QoS or Bandwidth Control.
• Netgear — Dynamic QoS under the main dashboard or Advanced Setup.
• Linksys — Media Prioritization or Smart Wi-Fi Tools.

If you cannot find anything labeled QoS, Bandwidth Control, or Traffic Management, your router may not support it. That scenario is covered later in this article.

The Most Important Setting — Set Bandwidth to 85 to 90 Percent of Your Real Speed

This single setting determines whether QoS actually fixes bufferbloat or does nothing at all. When enabling QoS, most routers ask you to enter your download and upload bandwidth. Do not enter the speed your ISP plan advertises. Instead, run a speed test first and note the actual measured values.

Then set your QoS bandwidth cap to 85 to 90 percent of those measured values. For example, if your speed test shows 100 Mbps download and 20 Mbps upload, set QoS to approximately 85–90 Mbps download and 17–18 Mbps upload.

This deliberate reduction is the entire mechanism that makes QoS work against bufferbloat. By capping throughput slightly below your connection’s true maximum, the router ensures that the bottleneck point is inside the router — where it has queue control — rather than at the ISP modem or DSLAM, where the router has no control at all. The router can now decide which packets go first instead of blindly pushing everything into an overflowing buffer downstream.

Why Setting QoS to 100 Percent Does Not Work

Many users enable QoS, enter their full connection speed, and see zero improvement. This happens because setting the limit to 100 percent means the router is still pushing data as fast as the connection allows. The bottleneck remains at the ISP side, and the router’s buffer still fills up with excess packets it cannot send fast enough. Nothing changes.

The 85 to 90 percent cap is not optional — it is the entire reason the fix works. You are sacrificing 10 to 15 percent of peak throughput in exchange for stable, low latency under load. On a 100 Mbps connection, this means you may see 85 Mbps during a speed test instead of 100 — but your video calls, gaming, and browsing will remain responsive even during heavy transfers. That tradeoff is almost always worth it.

How to Test If QoS Fixed Your Bufferbloat Score

After enabling QoS with the correct bandwidth cap, run the Waveform bufferbloat test again using the same wired setup you used before. Compare your new grade and latency numbers to your earlier baseline result.

If your grade improved from D or F to a B or A, the fix is working. Your loaded latency should now show a much smaller spike — ideally under 30 ms increase instead of the 200 to 800 ms spikes you saw before. If the grade did not improve, double-check that your bandwidth values are set to 85 to 90 percent of your real measured speed, not your plan speed. If the values are correct and the grade is still poor, your router’s QoS implementation may be too basic to handle bufferbloat effectively — which is where the next fix comes in.

Fix 2 — Enable SQM If Your Router Supports It

If basic QoS improved your grade but did not bring it down to an A, or if your router offers a more advanced option called SQM, this is the fix that produces the best results against bufferbloat. SQM is purpose-built for this exact problem and handles queue management far more effectively than traditional QoS.

What SQM Is and Why It Works Better Than Basic QoS

SQM stands for Smart Queue Management. While basic QoS simply caps bandwidth and applies rough traffic prioritization, SQM uses advanced algorithms — most commonly fq_codel or cake — that actively monitor every packet flow and make per-packet scheduling decisions in real time.

The difference is significant. Basic QoS prevents the buffer from overflowing by limiting total throughput, which works to a degree. SQM goes further — it identifies individual flows, ensures no single transfer can dominate the queue, and drops or delays excess packets intelligently before they pile up. The result is consistently low latency under load while still allowing bulk transfers to use as much bandwidth as possible.

In practical terms, SQM with the cake algorithm can keep your loaded latency increase under 5 to 10 ms, even with a large download and a video call running simultaneously. Basic QoS typically achieves 20 to 50 ms under the same conditions. For gaming and real-time voice or video, that difference is immediately noticeable. Packet loss caused by buffer overflow also drops dramatically because SQM manages the queue before it reaches that point.

Which Routers Have SQM Built In

SQM is not available on most stock consumer routers. The routers that include it either run custom-friendly firmware or come from manufacturers that specifically target network-aware users:

• ASUS routers running Merlin firmware — Merlin is a third-party firmware for ASUS routers that adds SQM support through its Adaptive QoS menu. It is the most popular option for users who want SQM without leaving the ASUS ecosystem.
• Routers running OpenWrt — OpenWrt includes full SQM support with the fq_codel and cake algorithms. Any router compatible with OpenWrt can gain SQM capability.
• IQrouter by Evenroute — A plug-and-play router designed specifically to eliminate bufferbloat. SQM is configured automatically.
• Ubiquiti EdgeRouter series — Supports fq_codel through its CLI and has SQM-like functionality built in.

If your router does not appear in any of these categories and runs stock firmware from TP-Link, Netgear, or Linksys, it most likely does not have true SQM. It may have basic QoS, which you should already have configured using Fix 1.

How to Enable SQM on ASUS Merlin Firmware

If you have an ASUS router that supports Merlin firmware, enabling SQM is straightforward:

1. Install Merlin firmware on your ASUS router if you have not already. Download it from the official Merlin firmware site for your specific model.
2. Log in to the router admin panel at 192.168.1.1.
3. Navigate to Adaptive QoS in the left sidebar and enable it.
4. Set the QoS type to Traditional QoS or Adaptive depending on your Merlin version. Some newer builds label the SQM-capable mode explicitly.
5. Enter your bandwidth values at 85 to 90 percent of your real measured speed — the same rule from Fix 1 applies here.
6. Under the queue discipline setting, select fq_codel if available. This is the algorithm that provides true SQM behavior.
7. Apply the settings and reboot the router.

After reboot, run the bufferbloat test again. You should see a noticeable improvement over basic QoS — especially in upload latency, which is typically the direction most affected by bufferbloat in home connections.

How to Enable SQM on OpenWrt

If your router already runs OpenWrt, enabling SQM requires installing one package and configuring it through the web interface:

1. Log in to the OpenWrt admin panel (LuCI) — usually at 192.168.1.1.
2. Navigate to System → Software and click Update lists.
3. Search for the package luci-app-sqm and install it. This installs both the SQM scripts and the graphical configuration page.
4. After installation, go to Network → SQM QoS in the menu.
5. Enable SQM on your WAN interface (usually eth0 or wan).
6. Set the download and upload speeds to 85 to 90 percent of your measured connection speed, entered in Kbps (for example, 85000 for 85 Mbps).
7. Under the Queue Discipline tab, select cake for both ingress and egress. The cake algorithm is the most effective option available on OpenWrt and handles both directions intelligently.
8. Under the Link Layer Adaptation tab, set the appropriate overhead for your connection type. For most standard Ethernet or cable connections, an overhead of 44 bytes with per-packet overhead selected works well.
9. Save and apply.

No reboot is needed — SQM takes effect immediately. Run your bufferbloat test right after applying. Most users see their grade jump to an A with loaded latency staying within single-digit milliseconds of their idle baseline.

Fix 3 — Flash Your Router With OpenWrt for Full SQM Support

If your router does not have SQM on its stock firmware and does not support Merlin, flashing OpenWrt is the most effective way to gain full bufferbloat control without buying a new device. It is not as risky as it sounds, but it requires some preparation.

How to Check If Your Router Supports OpenWrt

Before anything else, verify that your specific router model is compatible. OpenWrt maintains a comprehensive hardware database — the OpenWrt Table of Hardware — where you can search by brand, model, and version number. The version number matters. Two routers with the same model name but different hardware revisions can have completely different compatibility.

When you find your router in the table, check two things. First, confirm that a stable firmware build exists for it — not just a development snapshot. Second, check the flash memory and RAM specifications. Routers with less than 8 MB of flash or 64 MB of RAM will struggle to run OpenWrt with SQM effectively. The packages fit, but performance under load suffers because the router lacks the resources to run the cake or fq_codel algorithms without bottlenecking its own CPU.

What You Gain From OpenWrt SQM vs Stock Firmware QoS

The difference between stock QoS and OpenWrt SQM is not incremental — it is a generational leap in queue management. Stock firmware QoS on most consumer routers uses simple priority queuing or basic bandwidth throttling. It reduces bufferbloat partially, but it does not eliminate it.

OpenWrt SQM with the cake algorithm provides fair queuing across all flows, intelligent packet scheduling, and link-layer overhead compensation. This means your router accounts for protocol overhead that stock firmware ignores, resulting in more accurate bandwidth control and tighter latency management. Users who wonder why their WiFi feels slow even with fast speed test results often find that the root cause is exactly this kind of queue mismanagement that stock firmware cannot address.

In real-world terms, stock QoS might bring a Grade F down to a C or B. OpenWrt SQM consistently achieves Grade A on the same hardware with the same internet connection.

Is Flashing Your Router Worth the Risk

The risk is real but manageable. Flashing third-party firmware can brick your router if the process is interrupted or the wrong firmware file is used. However, most popular router models have well-documented flashing procedures and recovery methods.

The practical decision comes down to this: if your router supports OpenWrt, has adequate hardware specs, and your bufferbloat grade remains at C or worse after trying basic QoS, flashing is worth it. You gain a level of network control that consumer firmware simply does not offer. If your router is inexpensive or already a few years old, the risk-reward calculation favors flashing even more — the worst case is replacing a device you were already considering upgrading.

If flashing feels too risky, skip to Fix 4 below. It provides partial relief using only stock firmware tools.

Fix 4 — Limit Bandwidth on Specific Devices Causing the Problem

Sometimes the bufferbloat issue is not caused by general network congestion but by one specific device that consistently saturates the connection. A NAS running backups, a gaming console downloading updates in the background, or a smart TV streaming 4K content can each individually fill the buffer. In these cases, targeting that device directly can reduce bufferbloat without needing QoS or SQM at all.

How to Find Which Device Is Causing Bufferbloat on Your Network

Most routers display a traffic monitor or connected device list that shows bandwidth usage per device in real time. Log in to your router admin panel and look for a section labeled Traffic Monitor, Device List, or Network Map.

Watch the traffic values while the lag is happening. The device consuming the most bandwidth — particularly upload bandwidth — is usually the one triggering the buffer to fill. Upload-side bufferbloat is often worse than download-side because most home connections have significantly less upload capacity. A single device uploading at 15 Mbps on a 20 Mbps upload connection is enough to cause severe latency spikes for everyone else.

If your router does not show per-device traffic, temporarily disconnect suspected devices one at a time and rerun the bufferbloat test. When the grade improves after disconnecting a particular device, you have found the source.

How to Set Per-Device Bandwidth Limits on Your Router

Once you identify the problem device, you can limit its bandwidth directly. Most routers that offer QoS or Bandwidth Control also support per-device rules based on MAC address or IP address.

Navigate to your router’s QoS or Bandwidth Control section and create a new rule for the identified device. Set its maximum download and upload to a value that leaves enough headroom for the rest of the network. For example, if your connection delivers 100 Mbps download and 20 Mbps upload, capping the problem device at 60 Mbps download and 12 Mbps upload gives remaining devices sufficient breathing room.

This approach does not eliminate bufferbloat entirely — it reduces the severity by preventing a single device from monopolizing the connection. It is most effective as a temporary or supplementary measure alongside QoS or SQM rather than a standalone fix.

What to Do If Your Router Has No QoS or SQM Options at All

Some routers — particularly ISP-provided gateway devices and older consumer models — offer no QoS, no bandwidth control, and no path to SQM. If your router falls into this category, your options are limited but not nonexistent.

The most effective solution is to place a secondary router behind your ISP gateway and run QoS or SQM on that device instead. You set the ISP gateway to bridge mode or DMZ, then let the secondary router handle all traffic management. Even an inexpensive used router running OpenWrt with SQM will outperform an expensive ISP gateway with no queue management at all.

If replacing or adding hardware is not an option, the only remaining approach is manually limiting bandwidth usage at the application level. Most torrent clients, cloud backup tools, and download managers allow you to set speed caps within the application itself. Setting these caps to 70 to 80 percent of your connection speed provides some relief, though it depends on every user in the household cooperating — which is rarely reliable long-term.

The honest assessment: if your router has no QoS or SQM and cannot run OpenWrt, replacing it is the most practical path to eliminating bufferbloat permanently.

How to Confirm Bufferbloat Is Fully Fixed — Before and After Test

After applying any of the fixes above, you need a clear comparison to verify the improvement. Run the Waveform bufferbloat test under the same conditions you used for your initial baseline — wired connection, no other active traffic, full test completion.

Record three values from both your before and after tests:

• Bufferbloat grade — This is your primary success metric. Moving from D or F to A or B confirms the fix is working.
• Loaded latency (download direction) — Compare the latency spike during the download phase. A successful fix drops this from hundreds of milliseconds to under 20 to 30 ms.
• Loaded latency (upload direction) — Upload bufferbloat is often worse than download on home connections. Check that this value also dropped significantly.

If your grade reached A or B and both loaded latency values stay within 20 ms of your idle baseline, bufferbloat is effectively solved. Your video calls, gaming sessions, and browsing will remain stable even when someone else on the network is running a heavy transfer.

If the grade improved but is still at C, revisit your bandwidth cap setting. Try lowering it to 80 percent of your measured speed instead of 85 to 90. Some connections — particularly DSL and cable with variable sync rates — need a larger margin to keep the queue under control.

Frequently Asked Questions

Why does my internet lag when someone is downloading?

The download saturates your connection, and your router’s buffer fills with bulk transfer packets. Every other packet — including your browsing, voice, and gaming traffic — gets stuck behind that queue. This is bufferbloat, and it is a router-level queue management failure, not a bandwidth shortage.

Why do video calls drop when someone starts a download?

Video calls require consistent, low-latency packet delivery. When a large download floods the router buffer, video and audio packets are delayed by hundreds of milliseconds. The call application interprets this as connection loss and drops quality or disconnects entirely.

What is bufferbloat and how do I fix it?

Bufferbloat is excessive latency caused by oversized router buffers filling up during heavy network use. You fix it by enabling QoS or SQM on your router with bandwidth caps set to 85 to 90 percent of your real connection speed.

How do I test for bufferbloat on my router?

Use the Waveform bufferbloat test over a wired connection with no other network activity. It measures latency during full load and assigns a letter grade from A to F.

What does a bufferbloat grade D or F mean?

It means your loaded latency is spiking significantly — typically 200 ms or more above your idle baseline. Real-time applications like gaming, video calls, and voice chat will consistently degrade whenever the connection is under load.

Can I fix bufferbloat without buying a new router?

Yes, in most cases. If your router supports QoS, enabling it with correct bandwidth settings fixes or significantly reduces bufferbloat. If it supports SQM or can run OpenWrt, the fix is even more effective.

How do I enable QoS on my router to fix bufferbloat?

Log in to your router admin panel, locate the QoS or Traffic Management section, enable it, and set your download and upload bandwidth to 85 to 90 percent of your actual measured speed. Apply and retest.

What bandwidth percentage should I set for QoS to fix bufferbloat?

Set it to 85 to 90 percent of your real measured speed — not your ISP plan speed. This creates headroom that allows the router to manage the queue instead of letting packets pile up at the ISP bottleneck.

What is SQM and is it better than QoS for bufferbloat?

SQM uses advanced algorithms like fq_codel and cake to manage packet queues at a per-flow level. It is significantly more effective than basic QoS and consistently achieves Grade A bufferbloat scores where basic QoS may only reach B or C.

Why does my speed test look fine but gaming still lags?

Speed tests measure throughput on an idle connection. Gaming requires low latency under load. Bufferbloat causes latency to spike during active use while throughput remains unchanged — so your speed test looks perfect but gaming suffers.

Does enabling QoS slow down my internet speed?

It reduces peak throughput by 10 to 15 percent because you cap bandwidth below your maximum. However, real-world usability improves dramatically because latency stays stable. The tradeoff is almost always favorable.

Which routers support SQM for bufferbloat?

ASUS routers with Merlin firmware, any router running OpenWrt, IQrouter by Evenroute, and Ubiquiti EdgeRouter series all support SQM. Most stock consumer routers from TP-Link, Netgear, and Linksys do not.

How do I fix bufferbloat on an ASUS router?

Install Merlin firmware, enable Adaptive QoS, set bandwidth to 85 to 90 percent of your measured speed, and select fq_codel as the queue discipline if available. Retest with the Waveform tool to confirm improvement.

What is the Waveform bufferbloat test and how do I read it?

It is a browser-based test that measures latency while simultaneously loading your connection to full capacity. It assigns a letter grade from A to F based on how much your latency increases under load. Grade A means bufferbloat is not a problem. Grade D or F means severe bufferbloat requiring immediate action.

---

Bufferbloat is one of the few network problems that is almost entirely fixable at the router level without changing your ISP plan or upgrading your hardware. If your test showed a Grade D or F, enabling QoS with a correct bandwidth cap fixes the majority of cases. SQM takes it further and effectively eliminates the problem. If none of these options are available on your current router, a secondary router running OpenWrt with SQM is the most cost-effective permanent solution.

If you have applied SQM or QoS correctly, your bandwidth cap is at 85 to 90 percent, and your bufferbloat grade is still D or F, the issue may involve your ISP’s upstream equipment or an unusual connection type with variable sync rates. At that point, contact your ISP and specifically ask about latency under load on your line — not just speed. Provide them your before and after test results so they can investigate beyond standard throughput diagnostics.