Simple to follow. My latency went from +97 to +2. I may or may not play with up/down speeds but I doubt I will notice the difference. More importantly, I would never have found out about this pfsense optimization had it not been for another excellent Tom Lawrence video tutorial! THANK YOU.
Nice video! I would have been really interested in a brief explainer on what each setting does and why the given option was chosen (e.g. FQ_CODEL). I'm glad this brought bufferbloat to my attention, and I'd love a video going a bit more depth into what's going on.
what kinda sucks is that its really cumbersome to test various parts of each option - for example if you go for traffic shapers rather than limiters, some of them can add significant cpu overhead (FQ_CODEL as a limiter doesn't seem to do this). I'd love a video where someone actually tested the various overheads. also a lot of these videos focus exclusively on LAN/WAN connections (which most people have) but quite a few people run VPNs on the router now, and some even have all traffic running through VPNs. I've set up mine accordingly but it would also be nice to have a video focusing on those different kinds of configs. Some people run very high bandwidth applications too, which saturate networks (tor clients, torrent clients, open directory web servers / file servers) so it would be nice to have a best practice on how to ensure proper limiting of those (i did this through using the weights system and it seems to be working but i dont know if this is correct)
What a great video! I've implemented something similar with my Mikrotik router, and there are no more lags in the games. Thank you for sharing your knowledge with us.
I didn't think that this would be much of an issue for me since my dedicated firewall miniPC is relatively powerful. Went from a D (?!) to an A! Thanks for your great video and advice, Tom - as always! 👍👍
Needed this few months ago when Wide Open West forgot how to manage their network. Followed an older tutorial that helped some, but issue was WOWs end in the end. Best I could get was a C grade after limiting
I thought I already had this set up. Noticed some differences between your (the Netgate documented) method and what I had done, made the changes and it took me from an A to an A+. Thank you
Oh, so its basically what QOS settings do on other routers. Keeps your internet speed down JUST A BIT, so theres a little free bandwith for High priority things to jump through the Q instantly. Need to get this setup asap Thanks
Thanks for raising the issue! Having the consumer set up the limiter on their end is a sign of an ISP "cheaping out" - at least in my part of the world (Germany). My experience ranges form very well implemented (Telekom) to nothing at all (Vodafone Kabel - here a FW capable of setting up limiters was pretty much mandatory; at least in the past)
Vodafone is selling 32 Customers capacity on the same 2.5 gbit link. The fastest speed is 1 gbit that you can buy at the moment. Therefore a maximum of 2 people can have that at the same time. On DSL, the overbooking ratio was way lower. Therefore it is way more relyable with better latency. And the last thing worth knowing is the messed up peering Vodafone does. It is so much worse than the telekom ones, since they try to cheap out whenever they can. In Germany, we had over 6 months of almost broken cable internet because Vodafone canceled peering contracts.
@@andreas7944 exactly my experience after they shut down peering here in Berlin (before it was working quite well really well, even though I still needed a limiter because I would run into bufferbloat). Well, you only get what you pay for I guess!
Watch out for trash bytes in config files creating and deleting limiters. Same with changing unbound repeatedly. Sometimes you gotta manually fix things if you fiddle with settings too much. Never delete an interface without deleting firewall rules first. Trash leftover firewall rule bytes. If you use the layer 2 firewall be sure to have all your rules memorized. Lots of floating rules and layer 2 rule creation and deletion causes all of the rules to get mixed up randomly, and eventually I found it best to put my limiters and layer2 rules on non-quick rules. Last match rules and aliases are amazing. pfSense's rules are so awesome though 😅 If you can code your own firewall though, you are golden.
Big disclaimer for those on connections with unpredictable speeds!! My internet upload can vary between 35 and 50 mbps. I've tried every possible "buffer bloat" fix under the sun and nothing works if you can't set limits under which your speeds will not drop... Cable internet is a "shared medium" amongst everyone in your neighbourhood.
yep unfortunately you would have to set it ot 30 in this scenario and lose out on all that potential ;/ its not ideal but its probably the best you can do. maybe look into the scheduling options and see if you could toggle between them at various times, peak vs off peak
Totally correct. However there are other types of queues that do work better under such conditions. In regards to cable, the latency and the jitter is way worse compared to DSL. I would not recommend to use queues unless you have a real reason for that.
That's interesting; I assume you've tried CAKE autorate / adaptive bandwidth? Some Starlink & other satellite ISP users have used that with good success. CAKE is an alternative to FQ_CODEL, but it is unfortunately only on Linux-based routers (e.g., OpenWRT) and not quite yet on FreeBSD-based routers (e.g., OPNSense, etc). CAKE can be CPU-intensive to shape very high bandwidth (500Mbps+), though.
Not sure where I've gone wrong but my 1G fiber went from a C with 19 ms unloaded, +8 ms download, and +61 ms upload to a C with 17 ms unloaded, +8 ms download, and +154 ms upload. I'll have to spend some time testing different upload and download bandwidth values and see if I can improve things.
Curious. Did you try changing the Queue Length options in Download and Upload Limiter? I was trying to find ideal settings but the documentation is sparse.
Tom, as always, you and your videos are blessings for us novice IT adventurers. Thank you. I'm curious though, several of my more mobile devices use Wireguard to "phone home" when out and about. Would one need to make any accommodations for that when running the initial Bufferbloat test and/or setting up the CoDel Limiters? Running the Bufferbloat test with Wireguard activated on my laptop (connected via LAN) was almost 600 mb/s. Without Wireguard activated, my "gigabit" internet went back up to its usual 936 mg/s or so.
Thank you for a great tutorial and raising awareness of this issue. However, I think it might have been good to mention that in order to make this fully work in a IPv6 enabled environment, a second floating rule must be made for IPv6 (and not set the protocol family as IPv4+IPv6). As per the pfSense guide.
If you attempt to do this on with a IPv6 WAN address. The source in the floating rule can't be "WAN-address" because you probably don't NAT with IPv6, use "any" instead or if you have a fixed prefix, use that. (no I did not just spend 4 hours thinking my Pfsense was borked. What do you mean?)
This test was interesting. I have 2G/2G Internet. My devices on a 10G connection get A with latency 16/7/1 and down/up 2100/1900. My 1G test devices were C with latency 25/34/162 and down/up 529/924.
With NordVPN installed on a Netgate 1100, I get a D or F, latency of over 300 ms down, 200 up. With the VPN turned off, I get an A, latency is much lower. I may have to forget about using a VPN now.
Interesting, after playing with the settings for a while I found it did absolutely nothing for my download at +40. But it significantly changed up upload from +50 to +3.
Ran the test and got a B+...so naturally now I am wondering how/if I can do something like this in Unifi? I just got into Unifi two weeks ago, so I am still learning.
@@danielr3207 Agreed. We've had good success with Smart Queues. Our UDM-Pro with Smart Queues to 95% of upload / download (500 Mbps) gives us an A rating with 36ms / +24ms / + 0ms (unloaded, download, upload).
Quick question: How (or do) you alter this with dual/multiwan, with our without a gateway group? I just went through and did this on both WAN connections.
Per their documentation: For multiple WANs make a complete set of queues for each WAN and make a separate floating rule for each WAN. Ensure the rules do not match the source IP address(es) of the other WANs.
This is pretty cool. I have redundant WAN connection with two ISPs both with different bandwidth. I presume I would just created each one individually and assign it to each individual WAN interface even though they are in a gateway group? Or would I have to use the lowest common denominator and apply it to the gateway group.
My connection is 1000/50, originally my download latency was +45-55ms. After going through this tutorial and i am getting +200ms on the download. To fix this I change the WANDown queue length to 4250 and now i get +12ms. Question: Should I change anything in the tunables to compliment this change? in some of the forums they mention that if you increase the queue length, you should also change "net.inet.ip.dummynet.pipe_slot_limit" to something higher. Can you comment on this?
This is a problem on all routers. The steps in opnsense are almost identical, with a small difference: limiters (pfsense) are pipes (opnsense). And there's a separate UI section for the shaper rules - quite neat.
for some reason, I cannot get the rules to apply to and limit traffic unless I set the source to Any on each rule for ipv4 and ipv6. Any ideas why this is and if it is okay to do so?
Humm.. I followed a similar video from Chris Titus Tech (old video) and it added a s*** tone of floating filters.. not like your 2 filters. Perhaps i should redo them and follow yours?
@@LAWRENCESYSTEMS i have my upload speed set at about 10Mb lower than my max. Dwn = 900 Up = 115. i even played around with the amount from 1000 (standard) - 3000 (for higher dwnld).
can someone point me in the direction of where to obtain a pfSense router like what was used in this video? A proven make and model would be ideal for me as I don't have time to experiment. I have a 500/500 fiber ethernet network that I want to use it on. Thanks
I have 2 wan setup with load balancer group as a gateway. How would I properly setup Bufferbloat floating rules then? Should I be creating a floating rule for each Wan separately? Can anyone help me out with this?
Per the Netgate documentation for multiple WANs make a complete set of queues for each WAN and make a separate floating rule for each WAN. Ensure the rules do not match the source IP address(es) of the other WANs.
Would it make any difference with my 50mbps down and 10mbps up internet service that I'm sharing with two other houses? Pfsense in a proxmox vm on an HP sff I5 computer. I'm using Ubiquiti 2.4ghz wireless gear. I'm in the country and the internet provider choices are limited. The internet is unlimited data which all providers should switch to. Doesn't make since why some limit the data. Im on a fixed wireless service and a wireless home network to all three houses. Two houses over 1000 feet from office where the Rocket 2AC Prism with antenna is located. The third house is 75 feet from office with two nanostation 5AC locos connection.
It helps at all speeds, BUT you likely won't notice it consistently until max speed, since buffers aren't full until then, which triggers the bufferbloat problem
@@KleskReaver Thats what I thought because I have tried this over the last few years and dont see a difference with everyday tasks. I also only hit my peak speeds when downloading updates.
Because Rasool, the guy who invented it, says so on post ~720 in the playing with FQ_Codel forum. If you want to use the Codel queue management algorithm, you should pair it with QFQ or WFQ+ and source/destination masks. And none of that seems to apply to OPNSense, as fqcodel in OPNSense stinks without the source and destination masks. Maybe it is because of ipfw. I do kind of like OPNsense a bit more overall because of the security plugins. And they have a lot of pre-tuned netmap features. Suricata seems a bit more comprehensive in pfSense. But pfSense can be like a 18-speed tractor trailer while OPNsense an automatic with privileges.
Wao my latency changed from 34 to 6 you are the best for doing all these videos to help the community.
Simple to follow. My latency went from +97 to +2. I may or may not play with up/down speeds but I doubt I will notice the difference. More importantly, I would never have found out about this pfsense optimization had it not been for another excellent Tom Lawrence video tutorial! THANK YOU.
Went from F with +544ms down / +100ms up, to an A+ with 1ms up/down on a fiber connection. Crazy! Thanks so much!
Wow - was not aware of this issue (not a gamer). Implemented and went from a "B" grade with 45ms to "A+" with 1ms. Thank you!
Took me from A to A+ on my 920/110 line ... I do quite a bit of gaming so ideal for me. Thanks Tom!
Tommy is the best when it comes to Network optimization and explanation.
Nice video! I would have been really interested in a brief explainer on what each setting does and why the given option was chosen (e.g. FQ_CODEL). I'm glad this brought bufferbloat to my attention, and I'd love a video going a bit more depth into what's going on.
There is not really a "Brief" explanation which is why I referenced that other video.
what kinda sucks is that its really cumbersome to test various parts of each option - for example if you go for traffic shapers rather than limiters, some of them can add significant cpu overhead (FQ_CODEL as a limiter doesn't seem to do this). I'd love a video where someone actually tested the various overheads.
also a lot of these videos focus exclusively on LAN/WAN connections (which most people have) but quite a few people run VPNs on the router now, and some even have all traffic running through VPNs. I've set up mine accordingly but it would also be nice to have a video focusing on those different kinds of configs. Some people run very high bandwidth applications too, which saturate networks (tor clients, torrent clients, open directory web servers / file servers) so it would be nice to have a best practice on how to ensure proper limiting of those (i did this through using the weights system and it seems to be working but i dont know if this is correct)
completely agree. we need a 1h video explaining and comparing scenarios. you are the big one tom!
What a great video! I've implemented something similar with my Mikrotik router, and there are no more lags in the games.
Thank you for sharing your knowledge with us.
I didn't think that this would be much of an issue for me since my dedicated firewall miniPC is relatively powerful. Went from a D (?!) to an A! Thanks for your great video and advice, Tom - as always! 👍👍
Thank you for this! My Idle, Download and Upload latency went from 4/27/57 to 4/4/4.
Needed this few months ago when Wide Open West forgot how to manage their network.
Followed an older tutorial that helped some, but issue was WOWs end in the end. Best I could get was a C grade after limiting
I thought I already had this set up. Noticed some differences between your (the Netgate documented) method and what I had done, made the changes and it took me from an A to an A+. Thank you
Went from C to A. Thanks!
From C to A, awesome. I play games a tonne and am now noticing general responsiveness improvements too. Should of done this ages ago.
Oh, so its basically what QOS settings do on other routers.
Keeps your internet speed down JUST A BIT, so theres a little free bandwith for High priority things to jump through the Q instantly.
Need to get this setup asap
Thanks
Thanks for raising the issue! Having the consumer set up the limiter on their end is a sign of an ISP "cheaping out" - at least in my part of the world (Germany). My experience ranges form very well implemented (Telekom) to nothing at all (Vodafone Kabel - here a FW capable of setting up limiters was pretty much mandatory; at least in the past)
Vodafone is selling 32 Customers capacity on the same 2.5 gbit link. The fastest speed is 1 gbit that you can buy at the moment. Therefore a maximum of 2 people can have that at the same time. On DSL, the overbooking ratio was way lower. Therefore it is way more relyable with better latency. And the last thing worth knowing is the messed up peering Vodafone does. It is so much worse than the telekom ones, since they try to cheap out whenever they can. In Germany, we had over 6 months of almost broken cable internet because Vodafone canceled peering contracts.
@@andreas7944 exactly my experience after they shut down peering here in Berlin (before it was working quite well really well, even though I still needed a limiter because I would run into bufferbloat). Well, you only get what you pay for I guess!
hah so strange I just checked the documentation for that just yesterday! great timing, thanks!
C to an A for me as well, even on my lowly 50mbps business class circuit.
Thank you for this video! I learned a new thing and improved my network! You're a legend!
More videos like this, if you don't mind!
Pretty good response on my setup aswell! Thanks Tom
Started with A and ended with A followed details to a T. Dedicated 500/500 fiber
Thank you for this.
Happy it helped!
Very nice, Worked very well. Love it!!
Watch out for trash bytes in config files creating and deleting limiters. Same with changing unbound repeatedly. Sometimes you gotta manually fix things if you fiddle with settings too much.
Never delete an interface without deleting firewall rules first. Trash leftover firewall rule bytes.
If you use the layer 2 firewall be sure to have all your rules memorized. Lots of floating rules and layer 2 rule creation and deletion causes all of the rules to get mixed up randomly, and eventually I found it best to put my limiters and layer2 rules on non-quick rules. Last match rules and aliases are amazing. pfSense's rules are so awesome though 😅
If you can code your own firewall though, you are golden.
My Buffer bloat grade changed from an F to A+ and the latency changed from 63ms to 4ms download to 494ms to 5ms upload.
Always top quality content, thank you sir! 👍
You are a networking wizard!
I just couldn’t get this to work it just shut me out off it all but good video
Thanks for another great tip Tom, you're the best!
Big disclaimer for those on connections with unpredictable speeds!!
My internet upload can vary between 35 and 50 mbps. I've tried every possible "buffer bloat" fix under the sun and nothing works if you can't set limits under which your speeds will not drop...
Cable internet is a "shared medium" amongst everyone in your neighbourhood.
yep unfortunately you would have to set it ot 30 in this scenario and lose out on all that potential ;/ its not ideal but its probably the best you can do. maybe look into the scheduling options and see if you could toggle between them at various times, peak vs off peak
Totally correct. However there are other types of queues that do work better under such conditions. In regards to cable, the latency and the jitter is way worse compared to DSL. I would not recommend to use queues unless you have a real reason for that.
That's interesting; I assume you've tried CAKE autorate / adaptive bandwidth? Some Starlink & other satellite ISP users have used that with good success. CAKE is an alternative to FQ_CODEL, but it is unfortunately only on Linux-based routers (e.g., OpenWRT) and not quite yet on FreeBSD-based routers (e.g., OPNSense, etc). CAKE can be CPU-intensive to shape very high bandwidth (500Mbps+), though.
@@ikjadoon Thanks for the info. Unfortunately a no-go for me running pfsense.
@@Clarence-Homelab Ah, apologies, yes. Understandable.
Not sure where I've gone wrong but my 1G fiber went from a C with 19 ms unloaded, +8 ms download, and +61 ms upload to a C with 17 ms unloaded, +8 ms download, and +154 ms upload. I'll have to spend some time testing different upload and download bandwidth values and see if I can improve things.
Curious. Did you try changing the Queue Length options in Download and Upload Limiter? I was trying to find ideal settings but the documentation is sparse.
@@samsampier7147 I did, I saw in the docs where it suggested 3000-5000 for higher bw connections so I tried 4000.
Thanks!
Thanks you!
Tom, as always, you and your videos are blessings for us novice IT adventurers. Thank you. I'm curious though, several of my more mobile devices use Wireguard to "phone home" when out and about. Would one need to make any accommodations for that when running the initial Bufferbloat test and/or setting up the CoDel Limiters? Running the Bufferbloat test with Wireguard activated on my laptop (connected via LAN) was almost 600 mb/s. Without Wireguard activated, my "gigabit" internet went back up to its usual 936 mg/s or so.
Thank you for a great tutorial and raising awareness of this issue. However, I think it might have been good to mention that in order to make this fully work in a IPv6 enabled environment, a second floating rule must be made for IPv6 (and not set the protocol family as IPv4+IPv6). As per the pfSense guide.
If you attempt to do this on with a IPv6 WAN address. The source in the floating rule can't be "WAN-address" because you probably don't NAT with IPv6, use "any" instead or if you have a fixed prefix, use that. (no I did not just spend 4 hours thinking my Pfsense was borked. What do you mean?)
This test was interesting. I have 2G/2G Internet. My devices on a 10G connection get A with latency 16/7/1 and down/up 2100/1900. My 1G test devices were C with latency 25/34/162 and down/up 529/924.
With NordVPN installed on a Netgate 1100, I get a D or F, latency of over 300 ms down, 200 up. With the VPN turned off, I get an A, latency is much lower. I may have to forget about using a VPN now.
Interesting, after playing with the settings for a while I found it did absolutely nothing for my download at +40. But it significantly changed up upload from +50 to +3.
Ran the test and got a B+...so naturally now I am wondering how/if I can do something like this in Unifi? I just got into Unifi two weeks ago, so I am still learning.
Smart Queues in UniFi will do a similar job.
@@danielr3207 Agreed. We've had good success with Smart Queues. Our UDM-Pro with Smart Queues to 95% of upload / download (500 Mbps) gives us an A rating with 36ms / +24ms / + 0ms (unloaded, download, upload).
Quick question: How (or do) you alter this with dual/multiwan, with our without a gateway group? I just went through and did this on both WAN connections.
Per their documentation: For multiple WANs make a complete set of queues for each WAN and make a separate floating rule for each WAN. Ensure the rules do not match the source IP address(es) of the other WANs.
@@LAWRENCESYSTEMS Thanks, I believe I did that. Nothing broke at least :)
This is pretty cool. I have redundant WAN connection with two ISPs both with different bandwidth. I presume I would just created each one individually and assign it to each individual WAN interface even though they are in a gateway group? Or would I have to use the lowest common denominator and apply it to the gateway group.
No, per the instructions on their site you would create a limiter for each WAN connection.
Can you make a video on what gateways we should add the rules or on what it makes sense? E.g vpn client gateway?
My connection is 1000/50, originally my download latency was +45-55ms. After going through this tutorial and i am getting +200ms on the download. To fix this I change the WANDown queue length to 4250 and now i get +12ms.
Question: Should I change anything in the tunables to compliment this change?
in some of the forums they mention that if you increase the queue length, you should also change "net.inet.ip.dummynet.pipe_slot_limit" to something higher. Can you comment on this?
380ms to 1 ms is a wild difference
Is this a problem in OPNsense - or rather, can i use the same steps in OPNsense? I'll try the test site when my neighborhood has gone to sleep.
This is a problem on all routers. The steps in opnsense are almost identical, with a small difference: limiters (pfsense) are pipes (opnsense). And there's a separate UI section for the shaper rules - quite neat.
I was literally about to start a thread with netgate about this. 😂
Have you reviewed the netnuma r3 gaming router?
Thanks for the quick guide!
How do you handle this with (privacy) VPN gateways?
No point because using those adds latency and you are at the mercy of their network issues.
for some reason, I cannot get the rules to apply to and limit traffic unless I set the source to Any on each rule for ipv4 and ipv6. Any ideas why this is and if it is okay to do so?
what program do i have to install?and how?idk how intall this,any help please?
If I want to apply it to the PIA VPN to, do I douplicate all, or is it enouge to just one more floating role?
Thx you ! C to A+!! :):
How do we install/ update adamone v3 to v4 please
Humm..
I followed a similar video from Chris Titus Tech (old video) and it added a s*** tone of floating filters.. not like your 2 filters. Perhaps i should redo them and follow yours?
Hey! Thanks! This definitely got rid of my latency BUT i now have 80% less upload speed. What happened?
Maybe you set the upload speed too low.
@@LAWRENCESYSTEMS i have my upload speed set at about 10Mb lower than my max. Dwn = 900 Up = 115. i even played around with the amount from 1000 (standard) - 3000 (for higher dwnld).
can someone point me in the direction of where to obtain a pfSense router like what was used in this video? A proven make and model would be ideal for me as I don't have time to experiment. I have a 500/500 fiber ethernet network that I want to use it on. Thanks
I used this on a Netgate 4200.
@@LAWRENCESYSTEMS thanks!
What if you're using pfatt supplicant branch which breaks the traffic shaper on the wan interface?
I don't havea solution for that, post in their forums.
I have 2 wan setup with load balancer group as a gateway. How would I properly setup Bufferbloat floating rules then? Should I be creating a floating rule for each Wan separately?
Can anyone help me out with this?
Per the Netgate documentation for multiple WANs make a complete set of queues for each WAN and make a separate floating rule for each WAN. Ensure the rules do not match the source IP address(es) of the other WANs.
i have 1000 ms download and 800 ms upload, do these ms levels fell in games ?
Would it make any difference with my 50mbps down and 10mbps up internet service that I'm sharing with two other houses? Pfsense in a proxmox vm on an HP sff I5 computer. I'm using Ubiquiti 2.4ghz wireless gear. I'm in the country and the internet provider choices are limited. The internet is unlimited data which all providers should switch to. Doesn't make since why some limit the data. Im on a fixed wireless service and a wireless home network to all three houses. Two houses over 1000 feet from office where the Rocket 2AC Prism with antenna is located. The third house is 75 feet from office with two nanostation 5AC locos connection.
Test if you are having an issue first.
I have openvpn tunnel, do I need to do it also to the vpn?
Is there a way to made a Video about bandwith guarantee for a Service (voip)?
Run the traffic shaping Wizard, choose the Voip option.
Is there a way with limiters without the wizzard?
@@mani_AT Not that I am aware of.
Thx 🙏🏽
How do you see your buffer bloat score? Is there link we use?
Yes, it's in the documentation I referenced
Is this only needed if you hit your max speed provided while gaming or on a voip call?
It helps at all speeds, BUT you likely won't notice it consistently until max speed, since buffers aren't full until then, which triggers the bufferbloat problem
@@KleskReaver Thats what I thought because I have tried this over the last few years and dont see a difference with everyday tasks. I also only hit my peak speeds when downloading updates.
wow.. i didn't even know this was a thing..
What if you have vlans?do i need to change the wan ?
This fixes buffer bloat between the WAN and internal subnets.
I was going to hire you to setup my stuff....Finally back in my house but the contact page guy says you dont do that now? called me a home user... wtf
Why is queue mgmt on Tail Drop and not CODEL?
Queue Management Algorithm is Taildrop and the scheduler is FQ_Codel
Because Rasool, the guy who invented it, says so on post ~720 in the playing with FQ_Codel forum.
If you want to use the Codel queue management algorithm, you should pair it with QFQ or WFQ+ and source/destination masks.
And none of that seems to apply to OPNSense, as fqcodel in OPNSense stinks without the source and destination masks. Maybe it is because of ipfw. I do kind of like OPNsense a bit more overall because of the security plugins. And they have a lot of pre-tuned netmap features.
Suricata seems a bit more comprehensive in pfSense. But pfSense can be like a 18-speed tractor trailer while OPNsense an automatic with privileges.
I followed this to the letter and it borked my internet. Different settings may be needed with 5G based internet.
and where can I check the delay?
I just googled bufferbloat test and got a handful of sites
this worsen my connection went from having A to -A and +5ms increase. weird.
I received an A+ on the bufferbloat test site without setting up any limiters, so I guess I don't need them.
If it is not broken, don't break it!
didn't notice a difference on or off
i tried this and it went from D to F :-(
No more in depth conversation about prioritizing different classes of traffic or creating multiple queues or even the all-important ack queue?
That is covered in the first 46 minutes of this video th-cam.com/video/rF46PNid1Mo/w-d-xo.htmlsi=FF8OgXmKqJJCrV13
I actually tried fq_codel for acks only in OPNSense 😂
These settings actually made my connection worse, I had an A to begin with so I was hoping it would make it a A+ but I fell to a B. Oh well.
num num num Cake
what worries me is it sounds like a..
Windows 7 problem,
even Linux needs to purge memory..
Easier Solution:
Purchase a TP-Link AC1750 or Asus RT-AC86U
Install OpenWRT
Enable SQM
piece_of_cake