Slow WiFi speeds - only for LAN to WAN

I thought I'd create a new post to avoid hijacking the one about WiFI 6

Yesterday I had my FTTP connection installed (Speeds of 500/75).
Using a wired connection I do indeed reach the headline speeds.
This is when wired either directly to the router or via one of the unmanaged gigabit switches.

However, when testing WiFi speeds I can only achieve 200 to 240Mbps down.
At this point you'd suggest it's just how it is, need better kit etc. etc.

But...I have 3 x TPLink EAP245v3's with a wired back haul.
The router is a LinkSys WRT1900ACS with OpenWRT (wireless radios are obviously off).

If I test LAN<>LAN speeds (WiFi to QNAP NAS using a Speedtest package), then I can reach over 800Mbps using my MacBook.

The key bit of info here is the traffic routing
WAP -> Gigabit switch -> Gigabit switch -> router -> Gigabit switch -> QNAP

When the traffic for my internal tests go via the router, I do not see the traffic in the real-time graphs...which is confusing the hell out of me. Surely the switched traffic would appear there?!

When I test LAN->WAN I definitely see the traffic (on both the lan port, br-lan and wan). Route:
WAP -> Gigabit switch -> Gigabit switch -> router -> ONT

I've swapped out the router for the TalkTalk one they provided, leaving my WAP's in play, same results (perfect on wired, slower on WiFi, fast to QNAP).

I'm at a loss as to how to proceed, so am open to suggestions / explanations.

Thanks

I’m going to cross-post and repeat myself here. Just to keep all my drivel in the one place! 🤣…

Throughput vs distance between device and AP can massively vary. That’s just the nature of the beast. An AP will have a distinct radiation pattern too, it will certainly not be the same in all directions. Generally speaking the closer you are the better the throughput, but not too close, and there will be more ideal ‘sweet spots’ too. Walls and other obstacles simply kill WiFi performance.

For example here are the patterns from the Ubiquiti access points. They are all remarkably different. Mounting orientation and indeed the material all have massive effects.

How much latency is wireless adding to the speedtests when testing to the NAS and testing to external servers, compared to a wired connection? How do the jitter figures compare?

In reply to a post by MilesR:

If I test LAN<>LAN speeds (WiFi to QNAP NAS using a Speedtest package), then I can reach over 800Mbps using my MacBook.
….

The key bit of info here is the traffic routing
WAP -> Gigabit switch -> Gigabit switch -> router -> Gigabit switch -> QNAP

I'm at a loss as to how to proceed, so am open to suggestions / explanations.

Thanks

Can you please test with a temporary setup as follows:

WAP -> router -> NAS

In reply to a post by Pheasant:

Can you please test with a temporary setup as follows:

WAP -> router -> NAS

Yes, I believe I can. I'll need to try that tomorrow...(pesky work getting in the way!)

By the way, the input is much appreciated.

The traffic graphs on the router only show LAN to WAN and WAN to LAN traffic not LAN to LAN

Thanks
Dan

In reply to a post by danielhyde:

The traffic graphs on the router only show LAN to WAN and WAN to LAN traffic not LAN to LAN

Thanks
Dan

Yeah, so it seems. (I will Google this, but)...Do you know if there is a way to enable that so that I can "prove" my internal speeds are "legit"?

Hi,
Just a thought do you have access to your AP via omada or directly in browser as i thought those TP link ones had a throughput monitor that monitors any traffic, I know my TP link gear mesures any traffic goring through the device you could potentially disable any other wifi devices to make the tests more accurate then you don't have random spikes.

Here is my TP-Link throughput monitor

Just a thought

In reply to a post by RR_The_IT_Guy:

Hi,
Just a thought do you have access to your AP via omada or directly in browser as i thought those TP link ones had a throughput monitor that monitors any traffic, I know my TP link gear mesures any traffic goring through the device you could potentially disable any other wifi devices to make the tests more accurate then you don't have random spikes.

Here is my TP-Link throughput monitor

Just a thought

Thanks, it's a good shout, but the Omada interface doesn't have bandwidth monitoring for the APs.
Only total traffic per AP.
https://ibb.co/RchJ9sV

Thanks, it's a good shout, but the Omada interface doesn't have bandwidth monitoring for the APs.
Only total traffic per AP.
https://ibb.co/RchJ9sV

Ah i understand, another thought, you said your nas is a QNAP do you know if there is any way you can monitor network traffic if you were to download a file off of it .
I know with synology I can go in my resource monitor and see the total live bandwidth usage on the nas itself then it gives more accurate results that way.

The only other option i can think of is two pc's with Continuous internet speed test tool which allows local server and speed test between two devices however it does require two devices, I have only used the windows version myself but there is a version for mac apparently .

Actually, the "Activity Monitor" on OSX does show Data received/sent per second.
Whilst running my internal speed test (using my NAS), I can see it's matching the Mb/s speeds I see.
So that re-enforces the view that my internal LAN<>LAN WiFi speeds are accurate.
i.e. speedtest page shows 730Mb/s download, activity monitor shows 92MB/s

More interesting though...I realised I hadn't as yet tried the SpeedTest.net app on my iPad Air (4th gen).
So far I've only tried it on my MBP (2015), iPhone 12Pro & a 2018 HP Windows laptop.

The iPad Air gets full LAN<>WAN headline speeds using the SpeedTest.net app!
This means it's something about my other devices with the WAPs and WAN access.

I will be trying what Pheasant suggested (linking one of the WAP's directly to the router to bypass the switches)...but with the iPad Air tests I suspect this won't yield much. We'll see.

With a Synology NAS, if you run Docker then you can launch a speedtest server app and test your devices throughout and latency, locally or wherever, just using a browser/HTML5

In reply to a post by Pheasant:

With a Synology NAS, if you run Docker then you can launch a speedtest server app and test your devices throughout and latency, locally or wherever, just using a browser/HTML5

This is what I already have 😀
I just obviously wasn’t explicit enough.

The results of that are why I questioned the difference in LAN and WAN speed test results

Example:
https://ibb.co/z65t5Gx

MilesR

This is much more common than most people believe. Many devices over 2 years old were not speccd to deliver more than 200Mb WiFi speeds which is not really surprising as most consumers did not have connection speeds exceeding 80Mb

Even now the main reason for wanting WiFi speeds of over 200Mb on a consumer device is to speed test to see if you can saturate a connection higher than 300Mb.

In practical usage the main consumer reason for wanting connection speeds over 80Mb is so many devices can use it at the same time without contention between them and for that you don't need Wifi speeds greater than 200Mb on any one device, ( except the router / AP itself). .So manufactures cut costs by using lower spec WiFi ( Except Apple where cost is not an issue due to price and profit margin)

In reply to a post by kitcat:

MilesR

This is much more common than most people believe. Many devices over 2 years old were not speccd to deliver more than 200Mb WiFi speeds which is not really surprising as most consumers did not have connection speeds exceeding 80Mb

Yeah, I get that...it just doesn't make sense to me that the same devices can achieve much higher speeds when connecting to my internal server, but when testing to the WAN it's slow.

In reply to a post by Pheasant:

With a Synology NAS, if you run Docker then you can launch a speedtest server app and test your devices throughout and latency, locally or wherever, just using a browser/HTML5

I think that only works if you have Intel processors in the NAS itself though, I have one of the Realtek processors which don't support docker unfortunately. I am considering a new NAS though as it's going to get very full, in 6 months time I have a few TB of data stored offsite that I have to store on my systems again, I know one thing, it's not going to be quick downloading it all.
I think i looked at a 4 bay option so i could reuse my existing NAS as archival or offsite backup.

I mean when i speedtest offsite i just use my samknows whitebox as its normally accurate, I mean if you email samknows they might give you one as long as you have already signed up and registered interest.

edit
Just remembered what the thread was about
Just another idea that popped into mind, have you tried to delete your macs wifi configuration files or potentially maybe booting windows on your mac either in bootcamp or just a bootable usb?

Edited by RR_The_IT_Guy (Fri 08-Oct-21 16:59:25)

Yes you are right; Synology it has to be running an x86 64-bit processor. Although Docker does work on other ARM based CPUs, Synology have restricted it. Mine is a 17 series.

In reply to a post by RR_The_IT_Guy:

In reply to a post by Pheasant:

With a Synology NAS, if you run Docker then you can launch a speedtest server app and test your devices throughout and latency, locally or wherever, just using a browser/HTML5

I think that only works if you have Intel processors in the NAS itself though, I have one of the Realtek processors which don't support docker unfortunately. I am considering a new NAS though as it's going to get very full, in 6 months time I have a few TB of data stored offsite that I have to store on my systems again, I know one thing, it's not going to be quick downloading it all.
I think i looked at a 4 bay option so i could reuse my existing NAS as archival or offsite backup.

I mean when i speedtest offsite i just use my samknows whitebox as its normally accurate, I mean if you email samknows they might give you one as long as you have already signed up and registered interest.

edit
Just remembered what the thread was about
Just another idea that popped into mind, have you tried to delete your macs wifi configuration files or potentially maybe booting windows on your mac either in bootcamp or just a bootable usb?

I got myself a new 6 bay one last year when my old one died (due to the Intel CPU bug).
Then a few months ago I bought 4 x 10TB drives and 2 SSDs (for the system and apps). I shouldn’t need new drives for a while!

It’s only local backups though for now, although the really important stuff is in cloud storage.

As for the WiFi testing…now I have proven it works on at least one device (the iPad Air) I’m a little less concerned. I just need to do a little more tweaking / investigations as to what works and what doesn’t.

Which iPad Air generation do you have?

Edit - if it’s the current 4th gen, then it’ll be running WiFi 6 (802.11ax) - so should easily fill your 500 meg pipe and then some! More like double.

Edited by Pheasant (Fri 08-Oct-21 18:35:59)

In reply to a post by Pheasant:

Which iPad Air generation do you have?

It's a 4th Gen.

Got it a couple of months ago when Currys did a guaranteed £150 trade in for an old iPad, so I dug out the 10 year old (!) one that was in a drawer and bagged a new 64GB Air for £349.

Thanks for confirming. Your iPhone 12 Pro should have near identical performance to the IPad, however I think you said in the other thread you were getting 440/400 down/up on the iPhone on LAN to LAN is that correct?

In reply to a post by Pheasant:

Thanks for confirming. Your iPhone 12 Pro should have near identical performance to the IPad, however I think you said in the other thread you were getting 440/400 down/up on the iPhone on LAN to LAN is that correct?

That’s right.

Where I’m sat right now it’s 440/440 down/up for the iPhone and 650/600 for the iPad LAN to LAN Speedtest.

iPhone is hitting 200/75 and the iPad is hitting 290/75 for LAN to WAN tests.

I've just been playing around with the WiFi settings.
Have completely separated the bands (previously had one SSiD that covered both bands and one that's 5Ghz).
On the 5Ghz one I've changed the width from Auto to force 80Mhz.
I've also reduced the power of two of the three APs.

Now, I've just run the tests again on my iPhone and hitting 400/73 down/up for LAN to WAN.

So that's interesting. Still doesn't explain for me why I see this discrepancy with LAN v. WAN though.

Your iPad is doing about the top whack of what I’ve seen on my 802.11AC based APs using an iPhone12. The PHY rate for this device on AC is 866 Mbps; so you’re probably not going to see much more than 650 Mbps in both directions on your current APs, thereabouts what you’re getting…

Now the iPhone is an interesting one, it’s running the same wireless spec in theory as the tablet but only achieving two thirds of the throughput. So to my mind it’s rate limited due to poor radio performance.. Moreover it’s antennas will not be as performant as the tablet, so it will be far more sensitive to ideal placement, position and distance with respect to the APs.

I’m also not sure if the phone is holding onto a weaker signal from another AP. You could check by shutting down the other APs, switching off all or as much of your other WiFi client gear and doing a single client to AP test to see if you can get better throughout. The fact your throughput increased as you lowered your AP power leads me to believe that is the case.

Edited by Pheasant (Fri 08-Oct-21 20:38:46)

In reply to a post by Pheasant:

Your iPad is doing about the top whack of what I’ve seen on my 802.11AC based APs using an iPhone12. The PHY rate for this device on AC is 866 Mbps; so you’re probably not going to see much more than 650 Mbps in both directions on your current APs, thereabouts what you’re getting…

Now the iPhone is an interesting one, it’s running the same wireless spec in theory as the tablet but only achieving two thirds of the throughput. So to my mind it’s rate limited due to poor radio performance.. Moreover it’s antennas will not be as performant as the tablet, so it will be far more sensitive to ideal placement, position and distance with respect to the APs.

I’m also not sure if the phone is holding onto a weaker signal from another AP. You could check by shutting down the other APs, switching off all or as much of your other WiFi client gear and doing a single client to AP test to see if you can get better throughout. The fact your throughput increased as you lowered your AP power leads me to believe that is the case.

Thanks Pheasant.

I think that after my changes last night (separation of SSID bands, channel width + power output adjustments) I'm now seeing more consistency in the WAN speedtest results. Specifically in the MacBook & iPads.
I'm less worried by speeds to the iPhones, just because it would be incredibly rare to need that much speed anyway. I'm getting at least 200Mbps on it, which is more than enough.

I still find it odd that I was seeing the inconsistency / discrepancy in results between LAN and WAN...but it's simply not worth the effort to work out

Fair play Miles. WiFi performance issues can be challenging to diagnose (and fully resolve) and a bit of a diminishing return.

As long as your key devices are getting decent enough throughput that’s all that matters. Couple of other things I’d mention…

1. On the 5 GHz band try and eliminate all ‘legacy’ 802.11N based clients that could potentially be causing the AP to drop performance on that band to maintain compatibility
2. Try and keep all ‘low performance’ clients on the 2.4 GHz band if you can. Put the devices with the highest bandwidth performance requirements on the 5 GHz band
3. Check for no overlapping channels, esp. with 80 MHz wide setups (but your controller should look after that)
4. Turn down the AP output power as much as possible - though you’ve already done that but you could experiment with different transmit powers and also think about the location of each AP and it’s relationship with other APs. You could look to tweak the power on individual AP to possibly improve their performance. You can do a ready reckoner radio survey with your MBP
5. Don’t know if TP-Link supports it, but you could also experiment with “minimum RSSI” on the AP, such that weak (and hence slow) connections get dropped by the AP. However this might impact usability with all devices. As you’ve already split the SSIDs across the bands then beam steering isn’t necessary.

In reply to a post by Pheasant:

Fair play Miles. WiFi performance issues can be challenging to diagnose (and fully resolve) and a bit of a diminishing return.

As long as your key devices are getting decent enough throughput that’s all that matters. Couple of other things I’d mention…

1. On the 5 GHz band try and eliminate all ‘legacy’ 802.11N based clients that could potentially be causing the AP to drop performance on that band to maintain compatibility
2. Try and keep all ‘low performance’ clients on the 2.4 GHz band if you can. Put the devices with the highest bandwidth performance requirements on the 5 GHz band
3. Check for no overlapping channels, esp. with 80 MHz wide setups (but your controller should look after that)
4. Turn down the AP output power as much as possible - though you’ve already done that but you could experiment with different transmit powers and also think about the location of each AP and it’s relationship with other APs. You could look to tweak the power on individual AP to possibly improve their performance. You can do a ready reckoner radio survey with your MBP
5. Don’t know if TP-Link supports it, but you could also experiment with “minimum RSSI” on the AP, such that weak (and hence slow) connections get dropped by the AP. However this might impact usability with all devices. As you’ve already split the SSIDs across the bands then beam steering isn’t necessary.

Thanks again Pheasant.
When I first set up the Omada controller last year I did play around with the RSSI values (to improve handover)...but never settled on 'good' values for each.
However, having now adjusted the power output of two of the APs I may revisit this.