Best MESH system for gigabit internet ?

Hi all,

I am upgrading from 400Mbps to 900Mbps on Wednesday [ can't wait :) ] and was just wondering if my current MESH system needs upgrading to well to make full use of the new speed. The NOVAs are a tri-band (IIRC) system and have no hard wired backhaul

We currently have 4 Tenda NOVA MW12s throughout the house (3 downstairs and 1 upstairs); the one furthest from the router, downstairs, which is about 10m away from the main node gets speedtest results of 250Mbps - and this unit is also hardwired (with Cat5e cable) to multple devices including everything in our home office at the bottom of the garden.(where WiFi reach is pretty much non-existent). From a WiFi perspective, mobile devices (by far not the most modern of phones) get speeds between 40Mbps and 80Mbps throughout the property.

Whilst 250Mbps at the furthest ends of the property is fairly good IMHO I can't help but think it can so be so much better especially once I move up to 900Mbps coming into the house.

So a couple of questions:

1) Should a MESH system be showing a 150Mbps drop-off over 10m ? And is that drop-off a constant i.e on a 900Mbps connection should the furthest node be getting 750Mbps ?

2) Will upgrading to 900Mbps lead to an improvement in the speeds our mobiles are getting over WiFi ?

3) On a fairly restrictive budget - let's say no more than £200 - are there any options would be better than the current NOVAs ? I see lots of references to AC... and AX..., MESH systems but am completely lost in the jargon to know if they are any better.than what I have already ?

Oh - and before anyone suggests it - Powerlines are not an option (been there tried that - and they were pretty useless due to the wiring quality in the house and the fact the house has 2 ring mains(?)). Neither, really, unfortunately is laying Ethernet cable throughout the house

the TP-Link Decos seems to get good reviews and they are a pretty good price depending on what ones you go for, but if any wi-fi will hit 1Gb/s is another thing. The ones you have got according to a review I have seen will do 867Mbps in theory on a 5Ghz.

To be honest a desktop computer is always best to connect via Ethernet and games consoles as well, Wi-fi should only be used for stuff that can't be connected to Ethernet.

Powerlines are certainly not an option, they are not great.

Have a look at the TP link Decos. I had a look at them myself, but the problem is they only have three ethernet prots.

To get the most out of 900 Mbps broadband over WiFi along the furthest distance, you will likely need WiFi 6 based WiFi mesh, but from PC Pro magazines reviews in the last 6 months, this doesn't mean the most expensive.

In reply to a post by zyborg47:

The ones you have got according to a review I have seen will do 867Mbps in theory on a 5Ghz.

With 802.11ac (a.k.a. "Wi-Fi 5") the raw bitrate is up to 867Mbps, but in my experience the best *actual* throughput you'll get is around half that, even under ideal conditions.

If you run multiple access points to the same router with Cat5e cabling, and the APs are on different channels, then you can have different users on each AP using 400Mbps simultaneously.

But really, for most people I don't think it's worth attempting to tune your wifi to go any faster. 400Mbps is 50MB per second. If you're downloading a large application or software update, at this speed the time to download is already insignificant compared to the time to install it.

Sadly the setup and build materials of the house make fully wired connections out of the question

From the sound of it the NOVAs are WiFi-5 and the next step up would be WiFi-6 ?

What improvements would Wi-Fi 6 give over Wi-Fi 5 ? And presumably (without the wired back haul) I would need to be looking at a tri-band Wi-Fi 6 system ?

As a general rule of thumb, would be correct to say that "AC" refers to Wi-Fi 5 and "AX" refers to Wi-Fi 6 ?

Also, as I said earlier, most of the mobile devices are fairly old - presumably they will still connect to a Wi-Fi 6 system (though may not see any noticeable improvement) ? Though the primary concern is maximising the speed between the ONT/router and the MESH node (which is currently being used as the wired AP for the equipment at the far end of the house and garden)

I've also just remembered that "line of sight" is actually a thing with WiFi (which might explain the 150Mbps drop off) - EDIT: OK - just tried moving the nodes to have perfect line of sight (and unless I needed to reboot the nodes) this made no real difference to speed test results.

The ISP I am moving to offers a MESH system which seems to be this - https://support.plume.com/hc/en-gb/articles/36000529... - [ not sure if it's the WiFi 6 version or not ] for an extra £7pm. Might be worth considering ?

Edited by AdamBrunt (Sun 01-Jan-23 13:25:48)

In reply to a post by AdamBrunt:

From the sound of it the NOVAs are WiFi-5 and the next step up would be WiFi-6 ?

WiFi 6 = AX
WiFi 5 = AC
WiFi 4 = N

Then there are various versions of each type, you see AC1200 and AC2400 for example, generally these numbers are the theoretical max speeds added together (so completely useless) but they give a clue as to the capability of the device. The AC1200 units have less antennas and less capability than AC2400 or AC3600.

In reply to a post by jchamier:

In reply to a post by AdamBrunt:

From the sound of it the NOVAs are WiFi-5 and the next step up would be WiFi-6 ?

WiFi 6 = AX
WiFi 5 = AC
WiFi 4 = N

Then there are various versions of each type, you see AC1200 and AC2400 for example, generally these numbers are the theoretical max speeds added together (so completely useless) but they give a clue as to the capability of the device. The AC1200 units have less antennas and less capability than AC2400 or AC3600.

Thanks - that makes a lot sense.

You would think someone with a 20 year career in IT really ought to know this stuff right ???

In reply to a post by AdamBrunt:

You would think someone with a 20 year career in IT really ought to know this stuff right ???

The amazing thing about IT as a profession is how diverse it is, comparable only to medicine

You may find the US-centric Small Net Builder website and its forum of interest, although US house construction is quite different and WiFi seems to cover many more square meters (sq feet) than in traditional UK house construction.

If you can find a copy of PCPro magazine issue 318 they reviewed the Deco X20 which was the best price/performance WiFi 6 mesh at £270 back then. (The netgear mesh was nearly £500). The review tested range and performance in various rooms of a large suburban UK house, so may be of use.

Edited by jchamier (Sun 01-Jan-23 14:57:15)

Will give the DX20 a look over but it appears to be only dual band [ possibly why it is relatively so cheap ]; not such if that is going to be much of a deal breaker.

I could just get it from Amazon, try it and send it back if it ends up being a backward step from my current NOVAs.

Dual-band WiFi 6 v Tri-band WiFi 5 .... ???

In reply to a post by AdamBrunt:

Dual-band WiFi 6 v Tri-band WiFi 5 .... ???

I think the third band helps if you have a lot of constantly running connections, if your whole family internet use is typically bursty (even streaming in 4K UHD is actually bursty) then you might find as the speeds go up you don't need the third band, it really depends on usage, and if there is sufficient WiFi channels free.

WiFi 6 appears to be better optimised on both the 5 and 2.4 GHz bands, so even if your devices are WiFi 5 (AC) then you can see higher aggregate throughput.

Hard to test without using a lot of laptops and test software to a multi-gigabit connected NAS that can saturate >900 Mbps on wired - hence why worth looking at reviews or other forums.

Edited by jchamier (Sun 01-Jan-23 15:33:50)

In reply to a post by AdamBrunt:

Sadly the setup and build materials of the house make fully wired connections out of the question

From the sound of it the NOVAs are WiFi-5 and the next step up would be WiFi-6 ?
[/quote
TP-Link Deco X20 AX1800 are Wifi 6, but they are around £180 for a three pack.

[url]https://www.amazon.co.uk/TP-Link-Next-Gen-Extenders-HomeCare-Antivirus/dp/B0041JNUAI?ref_=ast_sto_dp&th=1

In reply to a post by candlerb:

With 802.11ac (a.k.a. "Wi-Fi 5") the raw bitrate is up to 867Mbps, but in my experience the best *actual* throughput you'll get is around half that, even under ideal conditions.

If you run multiple access points to the same router with Cat5e cabling, and the APs are on different channels, then you can have different users on each AP using 400Mbps simultaneously.

But really, for most people I don't think it's worth attempting to tune your wifi to go any faster. 400Mbps is 50MB per second. If you're downloading a large application or software update, at this speed the time to download is already insignificant compared to the time to install it.

I don't know an awful lot about mesh wi-fi networks, I did set one up for someone months ago, and I know about wired backhaul and wireless backhaul.
Thankfully, I have never had the need for in this house and wifi is pretty constant around the building and I have never needed super-duper speeds for Wi-fi

The dual band mesh will generally halve the throughput.

As you say most of your client devices are a bit older I'm guessing most will give a max of 400Mbit/s regardless of a higher speed AP, you'll only make use of the gigabit connection on a wired device.

In general I'd recommend tri band mesh. I use TP link APs myself EAP245 which are only AC hardwired, but on good clients in the same room I can get 650Mbit/s in close proximity.

I've fitted Deco M5 to a couple of relatives houses, hardwired, and they achieve 500Mbit/s to capable clients, but only dual band.

Within your budget of £200 you could probably grab a 3 pack of Deco M9 which are tri band AC. Tri band AC are probably going to give you a reasonably good mesh backhaul and client throughput to AC clients.

I usually test internally using an OpenSpeedTest sever on a newish PC, hardwired to the router (if you Google it you can download a sever for most OS and docker, not the online checker), particularly as my broadband is slower than the WiFi can do.

In reply to a post by adrenalize_:

The dual band mesh will generally halve the throughput.

As you say most of your client devices are a bit older I'm guessing most will give a max of 400Mbit/s regardless of a higher speed AP, you'll only make use of the gigabit connection on a wired device.

In general I'd recommend tri band mesh. I use TP link APs myself EAP245 which are only AC hardwired, but on good clients in the same room I can get 650Mbit/s in close proximity.

I've fitted Deco M5 to a couple of relatives houses, hardwired, and they achieve 500Mbit/s to capable clients, but only dual band.

Within your budget of £200 you could probably grab a 3 pack of Deco M9 which are tri band AC. Tri band AC are probably going to give you a reasonably good mesh backhaul and client throughput to AC clients.

I usually test internally using an OpenSpeedTest sever on a newish PC, hardwired to the router (if you Google it you can download a sever for most OS and docker, not the online checker), particularly as my broadband is slower than the WiFi can do.

I already have a tri-band AC [ Tenda NOVA MW12 ].

I might hold off until later in the week - the installation is on Wednesday - and see what speeds I get with the current setup first.

Good shout on the OpenSpeedTest server; I have Portainer on a Rasp Pi on the network which should be able to handle it quite well

Ah, apologies I wasn't familiar with the Tenda specs - it is a tri band mesh although I note on the specs both of the 5GHz channels are 2 stream - i.e. 867Mbps PHY link speed max.

So in reality with a client (phone/tablet/laptop) that is also 2 stream capable sat right on top of the main Tenda AP you will only likely see ~650Mbps throughput due to the overheads of WiFi etc.

The Tenda nodes that backhaul over WiFi will only see the same max - and that's when they are close together.

Yes the link speed is going to drop off pretty quicky from that as you move more than even a few metres away and especially with walls etc in the way. Same for clients or the nodes in a mesh.

For most of your clients which are probably only 2 stream this is probably as good as it gets - I guess in your setup where a number of devices are then connected to the ethernet of one of the Tenda nodes (if I read that right) then they are only going to be as fast as the node can connect back to the main unit.

If that's the bit you want to improve you're going to have to spend a bit to upgrade to a mesh with a better backhaul (AX, 4 streams etc) and watch they aren't a big distance apart.

The speed the mobile devices get (which is already up to 80Mbps) is of secondary concern I think (though, obviously, any improvement would be a bonus).

The primary concern is the speed the furthest MESH node gets from the main node (they are about 5-10 metres apart). Currently this node is getting speeds of 250Mbps [ off a 400Mbps connection into the house ]. And you did it right - the furthest node supplies the internet connection (over wired ethernet) to about 7 other devices (and very occasionally mobile devices)

As I understood from other sources (when I got the Tenda) one of the 5Ghz channels is used exclusively for the backhaul if that makes a difference.

I am also aware that the Tenda system doesn't let you muck about with the frequency it uses (can't remember how it selects that but UI doesn't let you change it) which may also be affecting it ?

As I say, I will probably do some initial testing with the Tenda once I am upgraded - to see what the Tenda can do before buying a replacement. The interesting thing will be - should it need replacing - would dual-band AX outperform the Tenda tri-band AC ?

Incidently, I got OpenReachServer running on a Pi and weirdly, through a wired connection from a PC which is only separated by a network switch, it only reaches 180Mbps down but 890Mbps up. Very odd ??

Edited by AdamBrunt (Mon 02-Jan-23 11:06:17)

In reply to a post by AdamBrunt:

Incidently, I got OpenReachServer running on a Pi and weirdly, through a wired connection from a PC which is only separated by a network switch, it only reaches 180Mbps down but 890Mbps up. Very odd ??

Have you tried it with both the Pi and the PC wired into the switch?

What is "OpenReachServer"? Can't find it via Google. Do you mean "OpenSpeedTest"?

Anything which uses HTML is potentially limited by the browser, so the best speedtest of all is something CLI based, like iperf3.

Server side: iperf3 -s
Client side (upload test): iperf3 -c x.x.x.x # x.x.x.x is server IP
Client side (download test): iperf3 -R -c x.x.x.x

In reply to a post by candlerb:

In reply to a post by AdamBrunt:

Incidently, I got OpenReachServer running on a Pi and weirdly, through a wired connection from a PC which is only separated by a network switch, it only reaches 180Mbps down but 890Mbps up. Very odd ??

Have you tried it with both the Pi and the PC wired into the switch?

What is "OpenReachServer"? Can't find it via Google. Do you mean "OpenSpeedTest"?

Anything which uses HTML is potentially limited by the browser, so the best speedtest of all is something CLI based, like iperf3.

Server side: iperf3 -s
Client side (upload test): iperf3 -c x.x.x.x # x.x.x.x is server IP
Client side (download test): iperf3 -R -c x.x.x.x

That was with the Pi and PC wired into the switch.

No idea where OpenReachServer came from ??!!?? I meant OpenSpeedTeat

180 Mbps "down" meaning from the Pi to the PC? Sounds like the PC has some sort of anti-virus / content scanner active.

Such tools are generally quite happy to let viruses escape to other machines

Try booting the PC from an Ubuntu live USB stick and testing from that.

In reply to a post by candlerb:

Try booting the PC from an Ubuntu live USB stick and testing from that.

Assuming Windows 10 or later, using the Task Manager to show network throughput whilst downloading the 1GB file from Thinkbroadband (or something of at least this size from Microsoft, Google etc) is worthwhile.

Some modern laptops seem to have something against booting from USB

Edited by jchamier (Mon 02-Jan-23 16:48:50)

In reply to a post by jchamier:

Some modern laptops seem to have something against booting from USB

I've had at least one where the BIOS had to be adjusted to make booting from USB the first choice (otherwise it booted from HDD even if a USB was inserted).

In reply to a post by candlerb:

180 Mbps "down" meaning from the Pi to the PC? Sounds like the PC has some sort of anti-virus / content scanner active.

Such tools are generally quite happy to let viruses escape to other machines

Try booting the PC from an Ubuntu live USB stick and testing from that.

Must be something to do with that particular PC as another laptop, wired into the same switch, gets download speeds of 800Mbps+

So preliminary tests with the Tenda are not good

New set up is; Hey! 900/900 connection, Archer C7 router connected to the ONT and Tenda MW12 MESH connected to the C7

Laptop wired into C7 gives a speed test of 690Mbps [ not overly impressive considering it's meant to be 900Mbps ]
Laptop wired into main MW12 node gives a speed test of 680Mbps [ excellent considering the above ]
Laptop wired into furthest MW12 node give a speed test of 350Mbps [ disappointing - only 100Mbps faster than when it was connected to a 400Mbps ISP connection ]
All wired devices connected to the furthest MW12 node, again, get about 340/350Mbps

Wireless devices connected to the MW12 MESH give a speed test of 85+Mbps [ this is more than ample for what the devices are used for ]

So, overall, a bit disappointing that the shortest wired connection is nowhere near the full 900Mbps (is it possible this is due to the router being used ?) but more concerning is the speeds got from the furthest MW12 node.

Clearly, it's not a limitation on the the node itself (given the main node right next to the router gets pretty much the same speed as the router does) but more than likely a combination of the distance and WiFi 5.

Anyone got any ideas ?

What can't you acheive with 340mbps?

In reply to a post by broadband66:

What can't you acheive with 340mbps?

That's a good point.

It's more a case of wondering why it's only getting 340Mbps ? Or maybe the issue, now, is why the router is only getting 690Mbps on a 900Mbps connection ?

Welcome to the world of LAN optimisation.

First thing to do is rule out your laptop: try booting it with an Ubuntu live USB stick (to rule out Windows, anti-virus and NIC driver issues) and doing the test again. If it's a very old laptop, try a newer one.

Then you can try to rule out the router: you can try disconnecting the router and terminating the PPPoE connection directly on the laptop (create a PPPoE virtual interface).

However, I think things are pointing at the router right now.

The Archer C7 is discontinued. AFAICS it has a QCA9558 with a single core at 720MHz, which sounds pretty underpowered for a gigabit to me. (My old router is a Mikrotik hEX PoE which has a QCA9557 @ 800MHz; with IPv4 and FastTrack optimisations it can almost fill a gigabit, but with IPv6 only about 300Mbps)

You could try running OpenWrt on it, in case it's more efficient than the official firmware:
https://openwrt.org/toh/tp-link/archer_c7

Very odd. When I said much the same thing in another post you disagreed with me. Though I was at least arguing that a connection speed faster than the internet connection is rarely useful. So what do you really think?

In reply to a post by candlerb:

Welcome to the world of LAN optimisation.

In reply to a post by candlerb:

Then you can try to rule out the router: you can try disconnecting the router and terminating the PPPoE connection directly on the laptop (create a PPPoE virtual interface).

However, I think things are pointing at the router right now.

The first bit sounds interesting but I think you're right on the 2nd bit

In reply to a post by candlerb:

The Archer C7 is discontinued. AFAICS it has a QCA9558 with a single core at 720MHz, which sounds pretty underpowered for a gigabit to me. (My old router is a Mikrotik hEX PoE which has a QCA9557 @ 800MHz; with IPv4 and FastTrack optimisations it can almost fill a gigabit, but with IPv6 only about 300Mbps)

I assume https://www.amazon.co.uk/TP-Link-OneMeshTM-HomeShiel... would be a more than adequate replacement (assuming I don't intend to use it's WiFi capabilities)

In reply to a post by AdamBrunt:

I assume https://www.amazon.co.uk/TP-Link-OneMeshTM-HomeShiel... would be a more than adequate replacement (assuming I don't intend to use it's WiFi capabilities)

The only specification I can find for this model is "dual core CPU" (without saying what sort of cores, or what clock frequency). All I can say is it'll probably be better than what you have.

Since you don't need the wifi, this does open more options for routers. One is to build your own using a small PC with a two NICs and something like pfSense. Another is a dedicated router without wireless.

I use a Mikrotik RB4011 which is much more expensive than your proposed TP-Link, but has a quad-core 1.4GHz ARM32 processor. Eats a gigabit in its sleep

The newer RB5009 is more powerful again, although you are forced to use RouterOS v7 (which is almost certainly fine for home use now. It was a rewrite-from-scratch and there were early stability problems)

Yeah, that Mikrotik is way over budget.

For a wired-only option I think https://www.amazon.co.uk/dp/B08QTXNWZ1 might be overkill as well but am not certain. Though it could have its uses whilst I am still getting Sky's 400Mbps AND Hey's 900Mbps at the same time

Edited by AdamBrunt (Wed 04-Jan-23 17:53:26)

What speedtest are you using? Try speedtest.net - it should default to the Hey! test server, if it doesn't change to it.

This should give you your speed to Hey!, hence the fibre speed, and what the router is capable of. You should get 800Mbps+ up and down. any less, and I'd say the router is the issue. You can rule this out with a direct connection from a decent spec PC to the ONT.

Unfortunately Hey! only has one peer (AS61231) so the choice of routes out onto the internet is strictly limited, so real-life speeds may be affected. I haven't noticed this though.

Glad you got it installed at last!

In reply to a post by AndyPandy:

What speedtest are you using? Try speedtest.net - it should default to the Hey! test server, if it doesn't change to it.

This should give you your speed to Hey!, hence the fibre speed, and what the router is capable of. You should get 800Mbps+ up and down. any less, and I'd say the router is the issue. You can rule this out with a direct connection from a decent spec PC to the ONT.

Unfortunately Hey! only has one peer (AS61231) so the choice of routes out onto the internet is strictly limited, so real-life speeds may be affected. I haven't noticed this though.

So all the tests were done at speedtest.net - but in my case it defaulted to F&W Networks (which the installer said was who he worked for). I assume you are talking about the left icon under the speedometer [ which can't be changed ]; I don't see any Hey! option for the right icon (which only shows 10 options in total) ?

Edited by AdamBrunt (Wed 04-Jan-23 18:31:56)

Yeah, I'n not surprised by the speeds to be honest. The WiFi is mainly the weakest link here.

Even though your Tenda Mesh is Triband both 5GHz bands are only 2 streams - so max PHY connection speed of 867Mbps - thoughput will probably only ever be ~650Mbps on a speedtest when right next to each other. As soon as you move them 5-10m apart I would expect the speed to drop. You may also find if you have them in a "row" so main talks to "2" then "2" talks to "3" then the throughput will drop/halve. You my not be able to see this in the settings as I'm not familiar with them.

The devices only geting 85Mbps - hmm, are they on 2.4GHz which is slower - or really old? I'm guessing 2.4GHz.

Agree the C7 is a bit old and may not be able to route any faster on the WAN, especially if PPPoE is required for Hey Broadband.

Is the C7 doing anything other than connecting the ONT to Tenda MW12? It looks like the Tenda can do routing, either DHCP, PPPoE - have you tried that to see if you can get the wired WAN speed a bit quicker?

This won't help the slave nodes though - as they are still connecting over the same WiFi backhaul.

Unfortunately the only real options are to spend a few hundred quid on faster AX kit or wire the nodes. Even then as you may struggle to achieve 900Mbps on a slave node at longer distances.

In reply to a post by adrenalize_:

Yeah, I'n not surprised by the speeds to be honest. The WiFi is mainly the weakest link here.

Even though your Tenda Mesh is Triband both 5GHz bands are only 2 streams - so max PHY connection speed of 867Mbps - thoughput will probably only ever be ~650Mbps on a speedtest when right next to each other. As soon as you move them 5-10m apart I would expect the speed to drop. You may also find if you have them in a "row" so main talks to "2" then "2" talks to "3" then the throughput will drop/halve. You my not be able to see this in the settings as I'm not familiar with them

According to the Tenda app it looks like all the nodes are talking to the all the other nodes at the same time; there is no obvious "node 3 has to go through node 2 to get to the master node"

In reply to a post by adrenalize_:

The devices only geting 85Mbps - hmm, are they on 2.4GHz which is slower - or really old? I'm guessing 2.4GHz

Very weird - just retested with my phone which is definitely capable of 5GHz and now speedtest.net says 550Mbps when sitting right on top of the master node and 225Mbps when sitting on top of the furthest node; which is a heck of a lot better . I think the Tenda nodes are not very quick/smart in handing over control to over nodes so that probably doesn't help the tests.

In reply to a post by adrenalize_:

Agree the C7 is a bit old and may not be able to route any faster on the WAN, especially if PPPoE is required for Hey Broadband.

Is the C7 doing anything other than connecting the ONT to Tenda MW12? It looks like the Tenda can do routing, either DHCP, PPPoE - have you tried that to see if you can get the wired WAN speed a bit quicker?

AFAIK PPoE is required by Hey!Broadband

So, currently, the C7 is just (a) connecting the ONT to the MW12 (b) the DHCP controller (managing a very small reserved IP list) and (c) acting as a switch for a couple of, nearby, hard wired devices. Long term, it will also be in charge of the DNS settings for all the devices on the LAN. I am aware that the Tenda can do PPoE and DHCP (not sure about reserved IPs) but I don't think it can do custom DNS settings. It also only has 1 WAN port and 2 LAN ports on it which may be a concern at a later date.

In reply to a post by adrenalize_:

Unfortunately the only real options are to spend a few hundred quid on faster AX kit or wire the nodes. Even then as you may struggle to achieve 900Mbps on a slave node at longer distances.

Yeah, think I might get some Deco DX20s from Amazon - and just return if they don't improve things much.

If I was going to wire the nodes I would also run wire all the way to the wired devices (which are currently just wired to the furthest node) but ... it is certainly the last resort at the moment

TP-Link AX20 router and Deco X20 3 pack both on their way from Amazon for testing purposes.

If testing is good not sure how I am going to get it past the missus though LOL

In reply to a post by AdamBrunt:

Sadly the setup and build materials of the house make fully wired connections out of the question

From the sound of it the NOVAs are WiFi-5 and the next step up would be WiFi-6 ?

Unless it's a rented property then Cat6a is always an option, The question is only how much disruption and expense you are willing to put up with to achieve it.

Personally in the real world gigabit ethernet is a waste of time if your only connections are going to be a wireless backhauled WiFi mesh system.

Were you just using the website?

Try downloading their Windows app from their site: https://www.speedtest.net/apps/windows

Here is what it looks like default for me, Hey connection, Hey server (High Wycombe): https://imgur.com/tfqKkYZ

This was my result run just now: https://imgur.com/doti7ZS

(This is before my wife starts working mind!)

In reply to a post by AdamBrunt:

Yeah, that Mikrotik is way over budget.

For a wired-only option I think https://www.amazon.co.uk/dp/B08QTXNWZ1 might be overkill as well but am not certain. Though it could have its uses whilst I am still getting Sky's 400Mbps AND Hey's 900Mbps at the same time

I can confirm that this model can handle your speeds no issues and is a very robust unit.
I use one in one of my offices for a CityFibre 1 Gig connection.

Thanks Dan

Ah, OK.

Yes, I have always just used the website - never even knew they had an app

TP-Link clearly have some very weird model naming; you would assume the AX23 - https://www.amazon.co.uk/TP-Link-OneMeshTM-HomeShiel... - was a better piece of kit than the AX20 - https://www.amazon.co.uk/gp/product/B089YXY2YH - but it doesn't look like it to me.

Gone for the AX23 for testing purposes

In reply to a post by danielhyde:

In reply to a post by AdamBrunt:

Yeah, that Mikrotik is way over budget.

For a wired-only option I think https://www.amazon.co.uk/dp/B08QTXNWZ1 might be overkill as well but am not certain. Though it could have its uses whilst I am still getting Sky's 400Mbps AND Hey's 900Mbps at the same time

I can confirm that this model can handle your speeds no issues and is a very robust unit.
I use one in one of my offices for a CityFibre 1 Gig connection.

Thanks Dan

That's good to know. I may test that as well if the AX20 doesn't come up to scratch

Though I am now wondering why I ordered both the router and the Deco [ as they are both AX1800 and quad-core ] so the router is pretty much redundant; other than the Deco only has 2 LAN ports on it (one of which, on the master node would be connected to the ONT)

In reply to a post by AdamBrunt:

TP-Link clearly have some very weird model naming; you would assume the AX23 - https://www.amazon.co.uk/TP-Link-OneMeshTM-HomeShiel... - was a better piece of kit than the AX20 - https://www.amazon.co.uk/gp/product/B089YXY2YH - but it doesn't look like it to me.

Gone for the AX23 for testing purposes

Tp-link are a bit strange on their model naming, some are known by two names, very strange.

In reply to a post by AdamBrunt:

Though I am now wondering why I ordered both the router and the Deco [ as they are both AX1800 and quad-core ] so the router is pretty much redundant; other than the Deco only has 2 LAN ports on it (one of which, on the master node would be connected to the ONT)

The decos are routers, but only one is used as a router, which is to be seems a waste, but I presume it no cheaper to produce one that is just a mesh wi-fi.
I looked at the Deco and the lack of lan ports is a problem, sure you can use a switch, but then it is another plug, another box and more cables.

In reply to a post by zyborg47:

In reply to a post by AdamBrunt:

Though I am now wondering why I ordered both the router and the Deco [ as they are both AX1800 and quad-core ] so the router is pretty much redundant; other than the Deco only has 2 LAN ports on it (one of which, on the master node would be connected to the ONT)

The decos are routers, but only one is used as a router, which is to be seems a waste, but I presume it no cheaper to produce one that is just a mesh wi-fi.
I looked at the Deco and the lack of lan ports is a problem, sure you can use a switch, but then it is another plug, another box and more cables.

Indeed,

I will do some double checking, later, to see just how many ports I am currently using on my current router [ maybe I can get away with just the 1 [i]non-WAN port ]

Well thank heavens for Amazon returns

The Deco X20, weirdly, in my situation is worse on pretty much both levels:

As a PPoE router it was giving slower speeds than the Archer C7
As a MESH system, in bridge mode to the C7, it was giving barely better speeds than the MW12s and had half the range of the MW12s i.e a 2nd X20 node, in the same position as the problem MW12, node wasn't even connecting to the main X20 !!??!!

Just waiting on the AX20 to arrive to see if that is better than the C7

In reply to a post by jabuzzard:

Personally in the real world gigabit ethernet is a waste of time if your only connections are going to be a wireless backhauled WiFi mesh system.

Even with 160 MHz wide 5 GHz channels with AX and WPA3 ? I'd read elsewhere that real world throughput upto 1500 Mbps was now possible, maybe in line of sight of course.

Not every ISP offers speed tiers, Hey! at least have three options, but the company digging the other side of my town is a single product/single price outfit, for 900/900.

Edited by jchamier (Thu 05-Jan-23 17:29:53)

In reply to a post by jabuzzard:

Personally in the real world gigabit ethernet is a waste of time if your only connections are going to be a wireless backhauled WiFi mesh system.

Not when your new ISP is charging you HALF the price for gigabit that your old ISP was charging was 400Mb or what some ISPs are still charging for 76Mb !!!

My home mesh uses the Eero Pro 6 system and its rock solid.

Well the AX20 is certainly better than the C7

A laptop hardwired into the AX20 is getting speeds of almost the full 900Mb. Wifi devices sitting very close to the main MW12 node are also getting very good speeds and devices connected to the furthest node are occasionally topping out at 425Mbps (so that's a lot better as well).

I think it also means that's as good as it going to get for the time being (due to budget constraints) and whilst I think of a possible easy way of getting a wired connection through to the other end of the house

In reply to a post by neo_wales:

My home mesh uses the Eero Pro 6 system and its rock solid.

At £560 for a 3 pack, you would hope it would be

In reply to a post by AdamBrunt:

In reply to a post by neo_wales:

My home mesh uses the Eero Pro 6 system and its rock solid.

At £560 for a 3 pack, you would hope it would be

It is Neo, he gets the best.

In reply to a post by zyborg47:

It is Neo, he gets the best.

Certainly pricey, 'best' can be debated. I gather there is an even more expensive WiFi 6E mesh from Netgear.

Yeah, the Orbi 960 is (RRP) £1699 for a 3 pack!! Unreal!! Probably better spent on the gas bill keeping warm.

I have tested a 6E router and my only 6E client (Laptop with Intel AX210 card) did get impressive speed, although I struggled to test the full potential as needed a couple of other devices wired to the router and OpenSpeedTest servers to get over 1Gbps.

However the 6GHz band has even less range than 5GHz and did drop off very very quickly, it really is designed for high speed in the same room as a cable replacement (kind of what WiFi was designed for in the first place!!) so makes me wonder how useful it is for backhaul - I don'y know if that is how the Orbi use it - they may use the 2nd 5GHz for backhaul and 6GHz just for local devices.

Location of mesh nodes is the crucial thing really - if they are only geting an average signal, they are never going to achieve great speed.

I wasnt that surprised with the Deco X20 not performing well, whilst they are WiFi 6, they are only dual band, so you'd be halving the throughput to a 5GHz client (assuming 5GHz backhaul, whilst your MW12 are tri-band.

>400Mbps at he other end of the mesh sounds pretty good though given the WiFi5/AC kit you've got.

In reply to a post by adrenalize_:

However the 6GHz band has even less range than 5GHz and did drop off very very quickly,

More popular in North America where their house construction is different, a lot less brick walls, so WiFi gets through. I've seen similar in rented offices with partition walls.

The European radio frequency regulators (EU & UK & Switzerland etc) have only approved half the capacity on 6 GHz than in North America, but still enough for two 160 MHz wide channels. Given the amount of systems on 5 GHz already, this could provide quite an impressive back-haul link between nodes, for then distributing using 5 GHz or even 2.4 to devices in a room.

The price may have something to do with supply issues if any parts made in or shipped through China.

In reply to a post by jchamier:

In reply to a post by adrenalize_:

However the 6GHz band has even less range than 5GHz and did drop off very very quickly,

More popular in North America where their house construction is different, a lot less brick walls, so WiFi gets through. I've seen similar in rented offices with partition walls.

Absolutely.

1960s built UK houses with double brick interior walls really don't like WiFi at all.

To be clear you asked why people needed more than the speed they were getting via their ISP. I'm pretty sure that person was on a 80/20 connection.

"Though I was at least arguing that a connection speed faster than the internet connection is rarely useful. So what do you really think? "

I think:

More than 50mbps per device is probably a waste of time and money in a domestic situation but a family with 2 devices each being used concurrently and access to a server within the premises will require faster than 80mbps.

The OP is paying less for a faster speed. It's a no brainer even if the top speed can't be reached.

Edited by broadband66 (Fri 06-Jan-23 16:27:26)

In reply to a post by AndyPandy:

Were you just using the website?

Try downloading their Windows app from their site: https://www.speedtest.net/apps/windows

Here is what it looks like default for me, Hey connection, Hey server (High Wycombe): https://imgur.com/tfqKkYZ

This was my result run just now: https://imgur.com/doti7ZS

(This is before my wife starts working mind!)

Do you mind sharing your BQM for your HB connection? They're building around my area (and at the same time Netomnia seems to be building too) and I am curious what their residential connection latency profile looks like. Thanks!

Sure, here you go: https://www.thinkbroadband.com/broadband/monitoring/...

It doesn't look pretty, and something weird happens at 2am and 2pm. Not sure if it's a Hey issue, or just my router prioritising other traffic.

I think that's fine. Remember that the yellow line in BQM shows "maximum latency" - only 1 in 100 packets gets the latency (99th percentile). The average (dark blue) barely budges, and there's almost no packet loss.

Between 2pm and 3pm there is a very *slight* increase in the average latency. I would guess some machine on your network was doing either a large upload or download - backup to a cloud service perhaps?

Edited by candlerb (Tue 10-Jan-23 17:57:30)

In reply to a post by AndyPandy:

Sure, here you go: https://www.thinkbroadband.com/broadband/monitoring/...

It doesn't look pretty, and something weird happens at 2am and 2pm. Not sure if it's a Hey issue, or just my router prioritising other traffic.

Very kind of you to reply promptly! That looks good - much better than all the jitter you get on Virgin Media....

I was allowed to pre-order as of last week, and I see engineers putting cables in the ground on.a daily basis now.

Do you remember how long it roughly took for them to call arrange install from pre-order? Contractor yesterday reckons about 3 months as an estimate.

Thanks again for sharing!

In reply to a post by AndyPandy:

Sure, here you go: https://www.thinkbroadband.com/broadband/monitoring/...

It doesn't look pretty, and something weird happens at 2am and 2pm. Not sure if it's a Hey issue, or just my router prioritising other traffic.

I never even knew that was a thing.

Set one up for my connection now and will check it fairly regularly.

Edited by AdamBrunt (Tue 10-Jan-23 19:16:00)

The 2am/pm thing seems to be happening to other Hey! customers, so I don't think it's me or my router. It's not noticeable, just an oddity on the BQM.

In reply to a post by nidstigator:

In reply to a post by AndyPandy:

Sure, here you go: https://www.thinkbroadband.com/broadband/monitoring/...

It doesn't look pretty, and something weird happens at 2am and 2pm. Not sure if it's a Hey issue, or just my router prioritising other traffic.

Very kind of you to reply promptly! That looks good - much better than all the jitter you get on Virgin Media....

I was allowed to pre-order as of last week, and I see engineers putting cables in the ground on.a daily basis now.

Do you remember how long it roughly took for them to call arrange install from pre-order? Contractor yesterday reckons about 3 months as an estimate.

Thanks again for sharing!

I "registered my interest" probably a good year or more before it became available. On the day it went live they sent me an email, and I booked the install that day. The install was about a week later.

I probably got lucky getting in first, so very quick for me.

If they're still cabling up, just keep an eye on the pole serving your house (if you have one). You want to see a white box with a number on it at the top, and a black duct running up to it with a yellow tag saying it's fibre.

Once that's in place, it won't be long until you can order!

In reply to a post by AndyPandy:

In reply to a post by nidstigator:

In reply to a post by AndyPandy:

Sure, here you go: https://www.thinkbroadband.com/broadband/monitoring/...

It doesn't look pretty, and something weird happens at 2am and 2pm. Not sure if it's a Hey issue, or just my router prioritising other traffic.

Very kind of you to reply promptly! That looks good - much better than all the jitter you get on Virgin Media....

I was allowed to pre-order as of last week, and I see engineers putting cables in the ground on.a daily basis now.

Do you remember how long it roughly took for them to call arrange install from pre-order? Contractor yesterday reckons about 3 months as an estimate.

Thanks again for sharing!

I "registered my interest" probably a good year or more before it became available. On the day it went live they sent me an email, and I booked the install that day. The install was about a week later.

I probably got lucky getting in first, so very quick for me.

If they're still cabling up, just keep an eye on the pole serving your house (if you have one). You want to see a white box with a number on it at the top, and a black duct running up to it with a yellow tag saying it's fibre.

Once that's in place, it won't be long until you can order!

Does it only have to be served via poles? My street (in fact most of neighbourhood) has not telecom poles. My current (albeit unused) openreach copper I think is cabled through to the property via underground. Does this mean they would have to put up poles?

It depends, if they can run the fibre following the existing copper, they will.

If they can't for some reason, they'll put up poles. This is what they're busy doing in a village near me.

Keep an eye on this site to track their activity: https://bidb.uk/

In reply to a post by AndyPandy:

The 2am/pm thing seems to be happening to other Hey! customers, so I don't think it's me or my router. It's not noticeable, just an oddity on the BQM.

Might be being blind but ... I don't see any 2AM thing on mine: https://www.thinkbroadband.com/broadband/monitoring/...

But the two spikes, at 7AM and just after 10AM seem like weird outliers to otherwise pretty good graph

Edited by AdamBrunt (Wed 11-Jan-23 17:35:56)

That's interesting, your graph looks much better than mine overall as well!

I am never sure how usage affects stuff like latency but .... even though I work from home (software development) all week most of the coding is done on a local laptop with the occasional connection to the cloud; other than that it's just general surfing and/or gaming by 3/4 people and the occasional block of big downloads and/or IPTV

So I am not exactly maxing out the connection all day if you know what I mean

Edited by AdamBrunt (Wed 11-Jan-23 17:40:15)