What's new

RT-AC68U Ping Spikes and Buffer Bloat

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

LagIsPain

New Around Here
Router:
Asus RT-AC68U Revision C1

Firmware:
Asuswrt-Merlin 384.6

Bridge:
Linksys E4200

ISP:
WISP with
50 Mbps Down
15 Mbps Up

Modem:
Ubiquiti POE
POE-24-12W-G-WH
https://www.wifi-stock.com/details/...-poe-adapters-white-5-pack-poe-24-12w-5p.html

Setup:
POE (Modem) -> Asus AC68U (Router) -> Linksys E4200 (Bridge)

Issue:
I will be on a 5 GHz Channel that only I use, and another user will be on the 2.4 GHz Channel. If any other user starts to use bandwidth, the ping will spike to over 1000ms and can stay there for several seconds. These ping spikes are caused by any activity, and doesn't come close to saturating the 50 Mbps down. Even as little as half a Mbps of activity on the network causes spikes.

The spikes seem to be specifically from changes in traffic on the network. Some examples:

1. No traffic on the network, start a ping test and ping is stable, someone loads a Facebook feed (which consumes 1 to 2 Mbps) and ping spikes 500ms to 1000ms.

2. Start playing a 1080p Youtube Video on a desktop connected to the 2.4 GHz channel. Bandwidth shoots up to 10 Mbps down. Start playing another 1080p video at same time on another computer, bandwidth shoots up to 20 Mbps. Run a ping test as videos are playing and ping is stable. (Really weird thing, start another video while running ping test and network is already using 20 Mbps, bandwidth use will go up, but ping will usually not spike or only a short spike of 50 to 100ms.)

3. Start a ping test, begin a photo upload, upload uses 2 to 5 Mbps, ping skyrockets to over 1000ms and stays there for duration of upload.

I have tested the router with multiple QoS settings, as well as just shutting QoS off. No matter what QoS settings I set, I get the exact same behavior every time. I've also tried DD-WRT firmware, but it performs even worse than Merlin.

This appears to be an issue with how the router handles traffic, as the spikes are all cause by changes in traffic on the network and not by saturation of available bandwidth. Anyone know any possible fixes to this problem or is the Asus AC68U just a router with horrible buffer bloat?
 
I appreciate all the information but, and maybe it's just me, I don't have a clear picture of your network.
"POE (Modem) -> Asus AC68U (Router) -> Linksys E4200 (Bridge)"​
You mention WISP, PCs, and WiFi channels but I can't correlate with your setup diagram. Also, how and why do the POE devices come into this. You're checking network performance with ping but what are you pinging?

Did this ever work or just started? Is there a real issue like poor website response or just a ping performance concern?

Possible packet segmentation issue. Have you tried running ping using different size payloads in increments of 250 or 500, with no other traffic? What you want to do is see where packet segmentation occurs.
 
maybe from your ISP

you could try John's fork of Merlin to see if it is performing better.
I know what you mean and had it in the past too. Any load on the line and ping is going up incredible.
 
WISP is my ISP

I mention WiFi channels to make it clear the lag is not from too many devices connected to one channel. I have two routers and 4 wireless access points (two 2.4 GHz and two 5 GHz radios), all on different WiFi channels so that they do not interfere.

The POE is the modem that my router is attached to, for connecting to the LAN. Some modems have buffers as well, but I don't think the POE does (included the information in case someone else was familiar with the device)

A more detailed explanation of the current LAN:
1. POE (Modem) -> 2. Asus AC68U (Router) -> 3. Linksys E4200 (Bridge)

1. POE (Modem)
Equipment provided by WISP.

2. Asus AC68U (Router)
Main router connected to WAN via POE. Provides two wireless access points for clients.

3. Linksys E4200 (Bridge)
Second router is in Bridge Mode and is connected to the Asus AC68U. Provides an additional two wireless access points.

Did this ever work or just started? Is there a real issue like poor website response or just a ping performance concern?

The ping has been an issue since I first got the router. It came with DD-WRT, but I recently installed Merlin to see if that would help. I actually get almost twice the bandwidth with Merlin, but the exact same ping issue exists (though slightly less bad). For web-browsing and watching videos, it really isn't noticeable, but for activities that are ping sensitive (AKA gaming), it is a major problem.

Possible packet segmentation issue. Have you tried running ping using different size payloads in increments of 250 or 500, with no other traffic? What you want to do is see where packet segmentation occurs.

I have done multiple tests under many conditions, everything from with the network idle to with multiple clients and the bandwidth fully saturated. I have also contacted the ISP and had them test the connection from their end. Every test I do on http://www.dslreports.com/speedtest I get either a "D" or an "F" for buffer bloat, regardless of time of day, traffic on the LAN, or settings on the router.

The ping spikes are whenever a different client on the network begins using bandwidth, and then stabilizes. For example:

User 1 is watching Netflix and using 8 Mbps. User 2 (that would be me) starts a ping test and the ping remains stable. User 3 opens up a browser and loads Facebook (uses 1 to 2 Mbps for several seconds), this causes ping to spike for 1 second to 500ms to 1000ms and then drop back down. Even though both User 1 and User 3 are using bandwidth, ping stabilizes back to normal. After a few minutes of browsing Facebook, User 3 loads a new website, causing another ping spike.

In the above example, every user is on a different wireless AP, so the lag spikes are not from WiFi latency caused by switching between users. Also, the combined bandwidth of everyone on the network is still far below the 50 Mbps of the WISP. And lastly, even when there is considerable bandwidth being used on the network, lag spikes only happen when a client sends out a new request. If the bandwidth stays stable at 5 or 20 or even 40 Mbps, pings also stay stable.
 
maybe from your ISP

you could try John's fork of Merlin to see if it is performing better.
I know what you mean and had it in the past too. Any load on the line and ping is going up incredible.

Did John's fork solve your problem? I'll give it a try if I can't find a solution with the current firmware I am using, as flashing is always a hassle.
 
Did John's fork solve your problem? I'll give it a try if I can't find a solution with the current firmware I am using, as flashing is always a hassle.
have much faster line now, John's known to be very stable so worth a try. Make a factory reset afterwards and better load it with restauration tool or miniweb. They are not fully compatible as he uses older more stable software parts.
 

Similar threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top