Measuring WiFi Performance

[frichardson]

I was a little surprised that my router doesn't come with some kind of throughput measurement tool. I figured I could mount a disk on the router and measure throughput using "rsync" to copy random file over. Is there a standard way of measuring throughput to/from the router?

 

[ColinTaylor]

No, there's no "standard" way because throughput means different things to different people.

To me throughput would mean the speed of network traffic passing through the router, i.e. WAN to LAN or LAN to LAN. rsync'ing a file to a USB attached disk is not measuring throughput at all, but the combined speed of the disk, USB interface, protocol, CPU, etc.

EDIT: If you are interested in the raw network speed, rather than how fast you can write to network storage or a given protocol, then iperf is one of the most commonly used tools.

 

[roadwrenchtech]

Are you trying to see how much data is moving through the router at any given time or find its maximum transfer rates?

 

[frichardson]

To me throughput would mean the speed of network traffic passing through the router, i.e. WAN to LAN or LAN to LAN. rsync'ing a file to a USB attached disk is not measuring throughput at all, but the combined speed of the disk, USB interface, protocol, CPU, etc.

EDIT: If you are interested in the raw network speed, rather than how fast you can write to network storage or a given protocol, then iperf is one of the most commonly used tools.

I figured the disk write speed would be very fast relative to WIFI throughput speed, but perhaps I am wrong there.

I will take a look at iperf, thanks!

 

[frichardson]

Are you trying to see how much data is moving through the router at any given time or find its maximum transfer rates?

I'm trying to figure out the maximum transfer rate over WIFI. speedtest.net was great for initial rough numbers (and for combined ISP + router performance), but I'd like to see how my router is doing on it's own.

 

[L&LD]

Even a 200Mbps ISP connection may show faster throughput that most USB devices (even SSD's) that we could plug into most Asus routers today. This is not a good test at all.

If you have Gbps throughput from your ISP, then fast.com gives me a hint at the max/highest throughput possible for routed connections.

With paid-for 1Gbps up/down ISP speeds, fast.com shows a 'final' wired speeds of 1.1Gbps up and 1.4Gbps down, regularly. With peaks of 1.4Gbps down and 1.9Gbps up using an i7-6700 with 16GB of Ram. Those speeds don't get matched with lower specced computers running the same tests.

 

[doczenith1]

I've done plenty of wifi tests on my network using iperf3. I have a wired computer on my network running in server mode and use the Android Magic iperf app on my phone. I seem to get the best throughput using this client command line iperf3 -c x.x.x.x -O5 -i 5 -w 4M -t 60

My Pixel 2 XL would top out around 650 Mbps and my Pixel 3 XL around 670.

 

[frichardson]

670 Mbps (out of a theoretical 850) is very close to the max I was seeing - I wonder if this is what everyone gets over 80MHz AC. I upgraded my adapter on my laptop and enabled 160MHz on my router (the theoretical max is now 1750), but in order to measure the max speed I'll need to run iperf on my router because the router has 1GHz ether ports. Kind of funny if my wifi throughput ends up being faster than wired...

 

[ACwifiguy]

I’m able to get mid 600Mbps on average on newer iOS devices. Low 700’s if I am within 3 feet of the AP.

Check out this write up - https://www.duckware.com/tech/wifi-in-the-us.html#wifioveread - lots of details on WiFi overhead on section 3. Makes sense - the weak point is on the client side.

You are not going to get better than 600-700’s Mbps in the real world with 2x2 clients on 802.11ac on 80Mhz. There aren’t any home applications that are going to require close to that amount of bandwidth either.

 

[frichardson]

I’m able to get mid 600Mbps on average on newer iOS devices. Low 700’s if I am within 3 feet of the AP.

Check out this write up - https://www.duckware.com/tech/wifi-in-the-us.html#wifioveread - lots of details on WiFi overhead on section 3. Makes sense - the weak point is on the client side.

You are not going to get better than 600-700’s Mbps in the real world with 2x2 clients on 802.11ac on 80Mhz. There aren’t any home applications that are going to require close to that amount of bandwidth either.

Thank you for that pointer! Yeah, that all makes a lot of sense - I'm just getting caught up in the geek factor of having a new router lol =D

 

[umarmung]

670 Mbps (out of a theoretical 850) is very close to the max I was seeing - I wonder if this is what everyone gets over 80MHz AC. I upgraded my adapter on my laptop and enabled 160MHz on my router (the theoretical max is now 1750), but in order to measure the max speed I'll need to run iperf on my router because the router has 1GHz ether ports. Kind of funny if my wifi throughput ends up being faster than wired...

You should not run measurement tools on the router itself, since you would be impacting its forwarding or bridging performance.

You need one powerful wired client to run the iperf server and then run the iperf client on a WiFi device for a single client WiFi throughput test.

 

[frichardson]

You should not run measurement tools on the router itself, since you would be impacting its forwarding or bridging performance.

You need one powerful wired client to run the iperf server and then run the iperf client on a WiFi device for a single client WiFi throughput test.

Yeah, the problem is that my ethernet ports are 1gig and I'm trying to see how much I can go over 1gig with my 1700 160MHz connection. I think I can run iperf briefly to do that and then disable it.

 

[thiggins]

If you are trying to assess > 1 Gbps throughput, you'll need to use more than one iperf3 client, connected to different LAN ports.

You can start tests to the same server, then add up the results.