There are two diagrams in the above drawing. I’ve labelled them (1) and (2).
I have a 1 Gbps internet connection.
With setup (1), if I do a speedtest on the phone, the download speed is about 250Mbps.
With setup (2), if I connect the router directly to the OTN, the download speed on the phone is 800Mbps.
I have eliminated cables and the switch, I think. For setup (1), the laptop connected to the switch gets about 900Mbps download. I’ve used the cable between the switch and router to connect the laptop and switch.
For the phone, I used the Android Ookla speedtest app.
For setup (1), if I use a web-based speedtest eg speed.cloudflare.com, the download is 500Mbps. I didn’t test this with setup (2).
I’ve used the router in router mode and bridge mode. I’ve also replaced the router with an access point. The result is the same.
I’m not sure if the speedtest reading for setup (1) reflects the actual download speed. On the phone, I timed a download of a file and it worked out at about 450Mbps.
However, the like for like comparison of using Ookla speedtest on the phone for setups (1) and (2) is what I’d like to understand. Why are the speeds (250Mbps vs 800Mbps) so different? Is this expected or do I have a configuration issue?
CPU load is low. Remember, when I do speed test via laptop wired to BLUE switch (shown in diagram), the speed is fine. I also have another switch (for GREEN network, not shown) and the speed test (wired) is fine for that too. So IPFire is coping with 1Gbps internet connection.
MTU for RED, GREEN, and BLUE is 1500.
In summary, the slowdown in only when I’m doing a wireless speed test on BLUE. The wired speed tests on BLUE and GREEN are fine.
I had tried the router in normal and bridge mode. I’ve also just replaced the router with an access point. Same result in all three cases. Disabling 2.4Ghz made a very slight difference (250Mbps → 270Mbps) but that may be within margin of error.
However, your post made me look at the services running:
If I disable IPS, the download speed jumps from 250Mbps to 450Mbps. IPS is enabled on RED, BLUE and OpenVPN. Disabling on just BLUE makes no difference. So IPS is contributing to some of the slowdown.
Another point of interest is that I previously had 70Mbps internet. And speed tests always roughly showed that speed. IPS was enabled then. It’s only when I switched to 1Gbps internet that the slower recorded speeds happened.
The 250 and 800 Mbps speeds you discuss…are these consistent, repeatable speeds in each set up? Are you the same distance from the access point in all tests? Is network activity equally quiet on all tests?
My point is, I have seen such variability between two wireless speed test runs which turned out to be due to extremes in variability and not anything else.
In my case, I have installed IPFire in FriendlyElec NanoPi R4S.
I have 1GB symmetrical connectivity, and with the Squid, Suricata (RED/OpenVPN/WireGuard), OpenVPN, WireGuard, IP Address Blocklist, Location Block, DHCP Server, Dynamic DNS, and Response Policy Zones (RPZ) modules, I’m getting 400MB download speeds and 120MB upload speeds.
If I disable the protections, I get more, minus the 1GB.
I know my hardware isn’t powerful, and for what I use it for, it works perfectly, 24/7 at 5V2A, or 10W (less than an old-fashioned light bulb).
In your case, the problem is likely the hardware where you have IPFire installed, since, as mentioned, many factors influence it, such as the processor, active/passive network cards, system interruptions, etc.
Good point. Until I made the IPS changes, all measurements were from the same place. I redid the post IPS changes. The results varied by about 10%. The 250/800 difference is too big.
Thanks. I have 1Gbps symmetrical too. My IPFire is not the most powerful machine (Celeron J4105 CPU @ 1.50GHz, 8GB RAM) but it’s handling wired connections at about the advertised speed. It’s only the wireless speeds (via the WiFi router) that slow down but which also speed up (to about 60% of max) when IPS/Location Block/IP Address Blocklist are switched off.
In the interest of science I did some testing. I have had 1g fiber for about month. So far, so good. It has worked for me although I know from random testing and reddit that the speed varies.
So here we are a 1PM EST and my desktop (wired) gets about 600 down and about 400 up. My Pixel 9a using same test in chrome (cloudflare) was about 400 up and 300 down.
The more tests I did the closer the speeds became with up between 400-600 and down 300-400.
I turned off IPS and blocklists and did see a slight improvement in both, but not enough to make a statistical difference.
Other times when I have tested both up and down have been around 900/900.
Also make sure that Quality of Service is turned off when testing. And preferably test during off-peak times. I know 1pm EST may be off-peak for you, but you just don’t know how many people using your ISP are working from home, etc. Better to try right when you wake up or just before you go to bed (assuming your have a typical sleep schedule).
Yeah, I know it varies by time of day. It always (so far) seems to be as good or better than my old cable internet.
I have not been able to get good results lately with QOS, so I do not use it. Even though my connection speed and latency seem OK to me, Waveform Bufferbloat gives me a B with QOS turned off and worse or no better with it turned on.
I am happy with my setup, I just wanted to see if there was a great disparity between phone and pc.
I am using an HP T730 and as long as it keeps working with IPFire I will keep using it.
Wired will always be more consistent and closer to line speed than wireless. Depending on your wifi type (N, AC, etc), and how close you are to the access point, you could get close to wired speeds under ideal circumstances. Another potential place to look is at the ethernet cable. Is it the proper grade? Is it terminated correctly? Does it have any damage? Do you have one or more switches between your wired PC and IPFire or is it a straight shot on one cable? Etc.
I retested in a browser using speed.cloudflare.com/. For phone (WiFi 6), I get 700 (down) and 560 (up). Switching off IPS/blocklists/location made no difference.
Using the Ookla Speedtest Android app, I get 250 up and down. Switching off IPS/blocklists/location increased to 780 down and 660 up. When connecting the router directly to the OTN (by passing IPFire), I consistently get over 800 up and down.
The web-based version of the Ookla speed test (speedtest.net) gives similar results to cloudfare.
Desktop (wired) is consistently above 900 up and down regardless of IPS, etc settings.
After multiple tests, results vary ±10%.
It looks like there is something about the Android Ookla speedtest app that makes it slow down when IPFire is used with IPS, etc.
The web-based apps are consistently high. I’m settling for this, especially since my previous ISP provided only 70Mbps download and was more expensive than my current 1Gbps fibre!
