Testing UT and pfsense bufferbloat /stress testing

[Chrismal]

Tried some test on our network that has 100 down and 6 up.

1) 3 PCs Downloading a Torrent (Debian ISO)
2) 1 PC uploading a file using wetransfer
3) A phone streaming a 4k video from Youtube
4) A phone running speedtest.net
5) 2 command prompts continuously ping google (ping google.com -t)

Untangle using Bandwidth control kept the ping around 13ms once in a while spikes to 29ms and No packets drop.

Pfsense codel discipline enable and also tried hfsc traffic shaper, Pings around 24ms spike to 75ms and also had packet drop while pinging

So for my test maybe was not a great way to test but that was the best I could try out at the moment.

TO ME UNTANGLED PROVED WHAT A GREAT PRODUCT IT IS

[richb-hanover]

These are pretty impressive results using an ad hoc test procedure.

I would love to hear the results of using DSLReports Speedtest (www.dslreports.com/speedtest) or one of the better tests suggested on What to do about Bufferbloat? Thanks!

[Chrismal]

Hi richb-hanover
Thanks for your post, I tried out the DSLReports but unfortunately I keep getting F on upload for buferbloat, this is same case for pfsense i tried limiting my upload speed to 2mbps with same results. for some reason the qos upload I limited to 4mbps but keep getting higher burst
DSL.PNG

[Chrismal]

Something is wrong in the dslreport I tried it from other connections with other routers and we keep getting same results I contacted the ISP and they checked it out and also think something wrong is happening when it ping the servers, and since manual test keep great pings there must be something wrong in that site or the servers it is pinging

[Chrismal]

Here is a screenshot of my manual test

Test.png

[richb-hanover]

Something is wrong in the dslreport I tried it from other connections with other routers and we keep getting same results I contacted the ISP and they checked it out and also think something wrong is happening when it ping the servers, and since manual test keep great pings there must be something wrong in that site or the servers it is pinging

I doubt that there's a problem with DSLReports - their test suite has been extensively reviewed for accuracy and consistency by bufferbloat experts.

Here are a few thoughts about your test results:

* Your test was heavy on downloads (torrenting, streaming). What happens if you had those three PCs uploading instead? (Many routers do well with downloads, but have too much buffering when uploading.)

* Perhaps your router isn't properly handling the latency-test probe that DSLReports uses. Since web browsers cannot send ping packets, DSLReports uses short web-socket queries to measure latency.

* Do you know if Untangle prioritizes ping packets? That would explain why your test looks very good.

* It would also be helpful to share the URL for the results page - the graphs give fine-grained data about how much/when the latency occurred.

Finally, I want to say that, "If it ain't broke, don't fix it." That is, if you're able to do everything you want without any problems, then don't worry about any particular test result. Thanks!

[sharrisonUK]

Do you know if Untangle prioritizes ping packets?

It can be manipulated in QoS -

Screen Shot 2017-02-24 at 15.44.30.png

[Chrismal]

richb-hanover

cheek this out just did a test and I did not touch anything. maybe it was a problem from my ISP

http://www.dslreports.com/speedtest/11011550


dsl.PNG

[JasonJoel]

Here is what I get... Network -> Untangle in bridge mode -> pfSense. No QOS or bandwidth control configured (just whatever is configured out of the box).

I can get bufferbloat to A+ if I apply pfSense's traffic shaping wizard settings, but then my download caps out at 25% max speed. Haven't had time to mess with it more than that.

Capture.PNG

[richb-hanover]

It can be manipulated in QoS -

Screen Shot 2017-02-24 at 15.44.30.png

I think I see why your ping results were so good. When you prioritize ping packets in the QoS settings, they "go to the head of the queue", and get sent right away. This means that any ping packets aren't useful for measuring latency, because they're treated specially.

However, you also got a very good result (an A) from DSLReports the second time. Do you think that the first test was at a different time of day? (Some ISPs do suffer congestion internally at busy times that contributes to high latency, and there's not much you can do about it.)

But as I said last time, if things are working well for you, leave thing alone and enjoy your life! Cheers.