Worth buying a new router for QoS?

[Drystoner]

In my home network there is usually about 6 active devices. The download speed is 40 mbps and upload 5 mbps.

While gaming even with ethernet, I get major ping spikes. On average I usually get 30 to 40 ms depending on the server. But when there is more traffic I get spikes anywhere from 100 ms to 500 ms.

Would buying a third party router that offers QoS be worth it in this scenario?

I'm planning to allocate 4 mpbs down and 0.2 mpbs upload per device. It is limited bandwidth to work with, but that's the fastest offered in this area.

[[N]ebsun]

What router or device do you currently have ? chances are it probably already has QoS (even if there's no settings for it, it might be there already)

QoS also doesn't put a cap on bandwidth - it just prioritises certain packets that match some defined criteria (like voip packets, so you don't get interruption on voice calls)

[Drystoner]

İlk olarak Nebsun tarafından gönderildi:

What router or device do you currently have ? chances are it probably already has QoS (even if there's no settings for it, it might be there already)

QoS also doesn't put a cap on bandwidth - it just prioritises certain packets that match some defined criteria (like voip packets, so you don't get interruption on voice calls)

It's a BT Smarthub 1. My device is a desktop using a CAT 5e cable plugged in the router.

The QoS seems to be non-existent as when I run the bufferbloat test, when download is full I get ping spikes of 100 ms and when upload is full it jumps to 700 ms with 40ms of jitter.

I've seen some routers allow you to set bandwidth caps on each device and that's really what I need for consistent ping. Getting 30 ms then random spikes to 500 ms is not acceptable.

[Azza ☠]

İlk olarak Kim Wexler tarafından gönderildi:

İlk olarak Nebsun tarafından gönderildi:

What router or device do you currently have ? chances are it probably already has QoS (even if there's no settings for it, it might be there already)

QoS also doesn't put a cap on bandwidth - it just prioritises certain packets that match some defined criteria (like voip packets, so you don't get interruption on voice calls)

It's a BT Smarthub 1. My device is a desktop using a CAT 5e cable plugged in the router.

The QoS seems to be non-existent as when I run the bufferbloat test, when download is full I get ping spikes of 100 ms and when upload is full it jumps to 700 ms with 40ms of jitter.

I've seen some routers allow you to set bandwidth caps on each device and that's really what I need for consistent ping. Getting 30 ms then random spikes to 500 ms is not acceptable.

When nothing else is using the Internet, check your connection performance with:
https://www.speedtest.net/

From a PC connected to the router with Ethernet cable. Does the download and/or upload spike or have a smooth flow? Note the ping, download, and upload speeds. Repeat the test on one of the Wifi devices to compare them.

You should be able to login to your router admin via typing on the web-browser URL:
//192.168.1.254

Then entering your router's admin username and password. Check on the Advanced settings > Advanced wireless, for Wireless Mode. Is that currently set to MODE 1? Is the Separate bands enabled?

Don't change anything, but look around the Internet setup/settings and see if you can figure out the ISP connection type: full fibre (FTTP), VDSL (FTTC) or G.Fast

All this will be important to know to ensure a new router supports it and would actually make a difference.

[HypersleepyNaputunia]

you could try out ones at store and return them if it doesnt help

[Bad 💀 Motha]

With that small # of devices, you don't even need QoS. Here's what I used to do at my parents place back in the day.

First off, I was the one who managed our Modem, Router, basically anything do with with Home Theater or PC stuff, cause that's what I was into. My parents were PC geeks too but they were not that tech-savvy with the networking stuff.

So I set it all up so anything in my room could connect via Ethernet Cable to the Router. Their stuff I made Wireless. To apply a QoS like setting all I had to do was change the WiFi mode from N to B/G; while my PC was connected via CAT5E @ 100Mbps. All their devices were now limited to 54Mbps or less (depending on signal and such).


If you can't maintain less then 100ms ping on a gaming server that is 3000 miles away or less, then your ISP just sucks really is all there is to it.

But overall 40 Down + 5 up; it's time for a better ISP speed plan. For that many devices actively used at most times during the same time period; then you really should have at least 100 Down + 10 Up minimum and a Modem/Router that is no more then 2-3 years old.

Modems and Routers should just automatically get replaced every 3-5 years anyways. If you receive and/or rent your Modem/Router through your ISP, you can do a few things here. First off look at what your ISP requires specs-wise for a Modem, this is public info they have to release to customers. Going by that you can look up 3rd party brand modems and see what is compatible; your ISP should have a fully disclosed list of those as well. You can opt-out of rental fee and flat-out buy your own Modem and/or Router. If you like your ISP modem, fine. But you should at least have an ISP modem (most are an all-in-one; Modem+Router these days) with Bridge Mode. Which allows you to use one of the Wired Ports on the Modem to bridge over to your own stand-alone Router. Then within the ISP modem, disable the onboard Wireless features, just leaving the wired ports. Then use your own Router for anything Wireless. And buy a Router that is at least Wireless N + AC Dual or Tri-Band (WiFi-5); or even Wireless AX (aka WiFi-6). Any half-decent $70+ Router should have all the needed and more common features like Advanced Firewall config options, Port Forwarding, QoS, UPnP; etc.

Even when I had Verizon DSL back in like 1999-2002 I ensured to have whatever was a decent Router at that time; which was maybe Linksys which even back then, had options for QoS and Port Forwarding and such. These days it is TP-Link / Netgear / ASUS

[Crawl]

I have an Asus Rapture GT-AXE11000 and you can set bandwidth limits per device by selecting them from the connected device list. It also has a game mode which prioritizes gaming traffic. It's a $500 router and may be overkill for your needs, I would assume that some of their cheaper models would be running similar firmware though.

[Bad 💀 Motha]

^ There are plenty of Routers within the $200 range that easily can do all of that.

[Lord Flashheart]

Maybe a second hand quad port nic in an old machine with pfsense installed.
It depends what you have already. A basic single port nic would be fine if you have a switch with enough ports.

I have an old machine with proxmox OS with virtualised pfsense, truenas 2 x pihole, lanlcache, windows + others.

[76561198343548661]

Tp-Link routers allow to enter bandwidth limits through an option called Bandwidth Control .

[[☥] - CJ -]

Are we sure the problem isnt their Internet as a whole and not just the current router when the internet is being used by multiple devices?

Download should technically be fine with multiple devices, but its your upload that could be bottlenecked and causing the ping spikes.

[Drystoner]

İlk olarak Azza ☠ tarafından gönderildi:

İlk olarak Kim Wexler tarafından gönderildi:

It's a BT Smarthub 1. My device is a desktop using a CAT 5e cable plugged in the router.

The QoS seems to be non-existent as when I run the bufferbloat test, when download is full I get ping spikes of 100 ms and when upload is full it jumps to 700 ms with 40ms of jitter.

I've seen some routers allow you to set bandwidth caps on each device and that's really what I need for consistent ping. Getting 30 ms then random spikes to 500 ms is not acceptable.

When nothing else is using the Internet, check your connection performance with:
https://www.speedtest.net/

From a PC connected to the router with Ethernet cable. Does the download and/or upload spike or have a smooth flow? Note the ping, download, and upload speeds. Repeat the test on one of the Wifi devices to compare them.

You should be able to login to your router admin via typing on the web-browser URL:
//192.168.1.254

Then entering your router's admin username and password. Check on the Advanced settings > Advanced wireless, for Wireless Mode. Is that currently set to MODE 1? Is the Separate bands enabled?

Don't change anything, but look around the Internet setup/settings and see if you can figure out the ISP connection type: full fibre (FTTP), VDSL (FTTC) or G.Fast

All this will be important to know to ensure a new router supports it and would actually make a difference.

Speedtest is quite useless as it only gives unloaded results. If I download something on say my phone and run the test then the ping and other results will be more reflective of what I get. I also do ping -t commands on command prompt to check my ping consistency which if all devices are in use cause major spikes.

Similar results also shown on the wifi devices, with worse overall ping due to not being wired of course.

[Azza ☠]

İlk olarak Kim Wexler tarafından gönderildi:

İlk olarak Azza ☠ tarafından gönderildi:

When nothing else is using the Internet, check your connection performance with:
https://www.speedtest.net/

From a PC connected to the router with Ethernet cable. Does the download and/or upload spike or have a smooth flow? Note the ping, download, and upload speeds. Repeat the test on one of the Wifi devices to compare them.

You should be able to login to your router admin via typing on the web-browser URL:
//192.168.1.254

Then entering your router's admin username and password. Check on the Advanced settings > Advanced wireless, for Wireless Mode. Is that currently set to MODE 1? Is the Separate bands enabled?

Don't change anything, but look around the Internet setup/settings and see if you can figure out the ISP connection type: full fibre (FTTP), VDSL (FTTC) or G.Fast

All this will be important to know to ensure a new router supports it and would actually make a difference.

Speedtest is quite useless as it only gives unloaded results. If I download something on say my phone and run the test then the ping and other results will be more reflective of what I get. I also do ping -t commands on command prompt to check my ping consistency which if all devices are in use cause major spikes.

Similar results also shown on the wifi devices, with worse overall ping due to not being wired of course.

While the speed tests might seem basic and pointless, there's a reason for that.

Data packet lose can occur down the line from the ISP (Internet Service Provider) to the Router itself. You don't want that! If it's a high ping and/or spiking on the download/upload, you should address that. In most cases, you can even call up your ISP and ask for a line test. They could adjust the package sending and optimize a bit their end for you to avoid package lost. Each data package lost or corruption, means the device needs to request it again.

If you are on an old ISP plan such as ADSL, they might offer you better ones with more bandwidth like Fiber, plus that might even come with a new router to support it's higher data connection. Then include those new features, such as QoS as an additional bonus. Something to consider as well, depending on the price tag and what you are currently on.

For QoS, it would also perform a speed test in a way to work out the max down/up rates to be able to spit up and share amount devices without leading to bottleneck. If it's not always at max but fluxuates, QoS won't help and you still will get issues ever now and then. It will even have to do it more across Wifi devices, leading to waste data traffic / repeated recalls across the network to finally get the valid data packages.

Then also speed testing again across to a Wifi device will compare the two, showing how much data lost or dropouts is actually across the current Wifi signal. How much you could actually be able to gain and stabilize. To a point at least, there will alway be some lost across the Wifi.

If you wish to use the ping command on your end, then try the following on a wired connection:

ping www.google.com -f -l 1450

The -l number value refers to a packet size to test. In the command above it's sending a package size of 1450.

You might get a reply from the website, loss = 0%, or not, rather a statement "Packet needs to be fragmented". If there's a good reply with no lost, keep repeating that ping in lots of 10 additional packages.

So while it's replying good:

ping www.google.com -f -l 1460
ping www.google.com -f -l 1470
ping www.google.com -f -l 1480
(etc)

It will fail sooner or later, giving you "Packet needs to be fragmented". Then go back to the previous one that worked and increase that one just by 1 each time.

ping www.google.com -f -l 1471
ping www.google.com -f -l 1472
ping www.google.com -f -l 1473
(etc)

Till it breaks again. Note down the last data package size which works.

MTU value = MSS (that value you just tested) + IP header (20) + ICMP header (8)

So just add that up. MTU is the Maximum Transmission Unit.

For example: MTU = 1472+28=1500 bytes. MTU = 1500 (optimized with no data lost)

Under the router admin, you will likely see a MTU size. Does it match that calculated value? Sometimes that MTU size will be set to automatic (specially on the newer routers), yet it should still mention it's own calculated value. That's the way you check it on your own end, rather than asking the ISP to check it via their own. If it's set to automatic, they will use their own and contacting your ISP for a line test could improve that.

[Bad 💀 Motha]

And most SpeedTests often will pick a decent test server not that far from you. What you also want to focus on here is manually doing speedtests against servers that are much further away and see what kind of results you get. Such as 150 miles, 500 miles, 1000 miles away; etc. Alot of times when you join an online game; the server or p2p-host might be 1000-3000 miles away for all you know.

[Drystoner]

İlk olarak Azza ☠ tarafından gönderildi:

İlk olarak Kim Wexler tarafından gönderildi:

Speedtest is quite useless as it only gives unloaded results. If I download something on say my phone and run the test then the ping and other results will be more reflective of what I get. I also do ping -t commands on command prompt to check my ping consistency which if all devices are in use cause major spikes.

Similar results also shown on the wifi devices, with worse overall ping due to not being wired of course.

While the speed tests might seem basic and pointless, there's a reason for that.

Data packet lose can occur down the line from the ISP (Internet Service Provider) to the Router itself. You don't want that! If it's a high ping and/or spiking on the download/upload, you should address that. In most cases, you can even call up your ISP and ask for a line test. They could adjust the package sending and optimize a bit their end for you to avoid package lost. Each data package lost or corruption, means the device needs to request it again.

If you are on an old ISP plan such as ADSL, they might offer you better ones with more bandwidth like Fiber, plus that might even come with a new router to support it's higher data connection. Then include those new features, such as QoS as an additional bonus. Something to consider as well, depending on the price tag and what you are currently on.

For QoS, it would also perform a speed test in a way to work out the max down/up rates to be able to spit up and share amount devices without leading to bottleneck. If it's not always at max but fluxuates, QoS won't help and you still will get issues ever now and then. It will even have to do it more across Wifi devices, leading to waste data traffic / repeated recalls across the network to finally get the valid data packages.

Then also speed testing again across to a Wifi device will compare the two, showing how much data lost or dropouts is actually across the current Wifi signal. How much you could actually be able to gain and stabilize. To a point at least, there will alway be some lost across the Wifi.

If you wish to use the ping command on your end, then try the following on a wired connection:

ping www.google.com -f -l 1450

The -l number value refers to a packet size to test. In the command above it's sending a package size of 1450.

You might get a reply from the website, loss = 0%, or not, rather a statement "Packet needs to be fragmented". If there's a good reply with no lost, keep repeating that ping in lots of 10 additional packages.

So while it's replying good:

ping www.google.com -f -l 1460
ping www.google.com -f -l 1470
ping www.google.com -f -l 1480
(etc)

It will fail sooner or later, giving you "Packet needs to be fragmented". Then go back to the previous one that worked and increase that one just by 1 each time.

ping www.google.com -f -l 1471
ping www.google.com -f -l 1472
ping www.google.com -f -l 1473
(etc)

Till it breaks again. Note down the last data package size which works.

MTU value = MSS (that value you just tested) + IP header (20) + ICMP header (8)

So just add that up. MTU is the Maximum Transmission Unit.

For example: MTU = 1472+28=1500 bytes. MTU = 1500 (optimized with no data lost)

Under the router admin, you will likely see a MTU size. Does it match that calculated value? Sometimes that MTU size will be set to automatic (specially on the newer routers), yet it should still mention it's own calculated value. That's the way you check it on your own end, rather than asking the ISP to check it via their own. If it's set to automatic, they will use their own and contacting your ISP for a line test could improve that.

It's a VDSL and the fastest speed offered here is 40 mbps with around 8 mbps up maximum. Other ISP's have worse speeds on offer than BT.

There is no fibre at all here as OpenReach has yet do that in this area. Which could take until 2026.

I ran another test to see ping and packet loss with loaded vs unloaded network. Again suggests bufferbloat and not even my CAT5e cable directly plugged in the router resolves it.

I've been suggested getting a TP-Link modem to hook up with the Amazon Eero mesh router as that has SQM CAKE which is good for those with low amounts of bandwidth still using DSL. It prioritises gaming and conferencing.

Here are my results:
https://imgur.com/a/7wMDZ7l

[Drystoner]

İlk olarak Bad 💀 Motha tarafından gönderildi:

And most SpeedTests often will pick a decent test server not that far from you. What you also want to focus on here is manually doing speedtests against servers that are much further away and see what kind of results you get. Such as 150 miles, 500 miles, 1000 miles away; etc. Alot of times when you join an online game; the server or p2p-host might be 1000-3000 miles away for all you know.

And again it's not very useful for gaming as traffic in the router changes all the time unless you're alone.

That's why I find bufferbloat test more useful and doing ping loops on command prompt.