Work from home surge may work in Wi-Fi 6's favour, reckons analyst house The COVID-19 pandemic looks set to have both short and long-term implications for the tech industry, themselves a reflection of changes in the established working culture. A new white paper from ABI Research highlights the growing need for modern home Wi-Fi kit to cope with an increasingly remote workforce. The paper points …
alongside IPv6 for the ISP's to ignore.
How many ISP's support IPv6 with their cheapo crap routers that we get to use?
What incentive, apart from a huge increase in monthly charges will there for them to give us WiFi 6?
TBH, I'd rather have a decent configurable firewall in their router before WiFi 6.
The main issue is the ISP feeding the box with WiFi.
This is frankly PR to sell their kit which will make little difference to people on 100Mbps broadband and none to people on 20 Mbps or less Broadband.
We only use WiFi for the phones, tablets and ereaders. Everything else is on 1 gigibit ethernet via cat 5 or Cat5e, mainly for the server and printer/scanner access. It makes little difference to the use of the 20 Mbps down, 5 Mbps up VDSL. So called "Fibre Broadband". Yeah, well the exchange has fibre, insist all the ISPs.
kit which will make little difference to people on 100Mbps broadband and none to people on 20 Mbps or less Broadband
I came here to make a similar point. Except in very crowded situations (I suppose you might get that in a block of flats) anything "g" and beyond will be fast enough for remote working that the bottleneck is usually the WAN, not the LAN - especially the uplink speed.
We're coping reasonably well, believe it or not, on ADSL2+ which syncs around 8Mbps/1Mbps, even in these days of video conferencing, home working and home schooling, but I have long had another use case for faster uplink speed so I'm beginning to look at FTTC, which is the only other real option where we live.
There seem to be two standard tiers, marketed as "up to 40Mbps" and "up to 80Mbps". We are relatively close to the cabinet so I'd hope to get close to those figures (doing the online check suggests we should get a minimum of 68Mbps on the top tier), but I'm more interested in the uplink speed, which seems to be 10Mbps or 20Mbps.
68Mbps down is much more than we will need in the foreseeable, but I have several plans which could make good use of as much uplink speed as can be made available. I've yet to hear back from the enquiries I've made, but there must be a way to rebalance the line - does anyone know of an ISP which would be happy to sell me an "up to 60/40" product rather than an "up to 80/20"?
M.
Ok, thanks.
I realise because of the way it works that it wouldn't be possible to rebalance completely the other way (wouldn't want to anyway), but I thought the boundary was movable to an extent. Oh well. FTTP and similar technologies are years away here, and out of my (home) budget at the moment anyway, so it'll be interesting to see what difference we notice with FTTC.
M.
> Except in very crowded situations (I suppose you might get that in a block of flats) anything "g" and beyond
The best solution in a crowded radio environment is still: move to the 5GHz band where there are many more channels than on the 2.4GHz band, plus the signal is attenuated more by walls etc. so less signal noise from the neighbours.
>but I'm more interested in the uplink speed, which seems to be 10Mbps or 20Mbps.
Yes the uplink speed is typically 25% of the download speed.
However, currently, I typically have ~20 concurrent remote RDS users using a 40/10 service and the uplink is showing the uplink only being loaded to 1~2Mbps, and none are complaining that things are slow etc. so I'm happy. (The lockdown happened before we were able to relocate the servers to a new facility with a 100/100 FTTP service...)
Okay, as they are remote, they aren't using the server uplink for Zoom etc.
The best solution in a crowded radio environment is still: move to the 5GHz band
Which is exactly what an upgrade gives you. From 802.11ac onward, 5GHz band support is required, and your devices will choose to use the 5GHz WiFi signal whenever available, and 2.4GHz only when necessary. Of course if you are able to shut of the 2.4GHz band on your router entirely, good on ya!
>Which is exactly what an upgrade gives you.
5GHz capable kit isn't new, just that it has taken time for it filter down to consumer-grade product.
Yes, it was good that support for 5GHz was mandated in the 802.11ac release, so it is slightly easier to buy a 5GHz capable router.
However, if you really want a fully capable dual-band router ie. circa 2007 "enterprise grade" kit, then you are still having to step away from both the stuff most ISP give away and the high street..
The best solution in a crowded radio environment is still: move to the 5GHz band where there are many more channels than on the 2.4GHz band
In theory, yes, but in practice I note that most "consumer" gear (that is, typical access points) does not seem to implement the Dynamic Frequency Selection and Power Control necessary to operate on the "top" eight channels and is instead restricted to the first four, which makes the total availability of space pretty much the same as at 2.4GHz, albeit with fewer users at the moment.
I typically have ~20 concurrent remote RDS users using a 40/10 service and the uplink is showing the uplink only being loaded to 1~2Mbps, and none are complaining that things are slow etc.
That's interesting because I am occasionally remoting-in to certain key devices using RDP over bog standard ADSL and the 1Mbps uplink speed of the originating end is absolutely a bottleneck with complex screens taking perhaps three or four seconds to transmit and even moving a window on a simple static desktop needs patience. Using the same endpoints on the LAN makes for an almost seamless experience.
Edit to add: though it isn't the RDP thing that's interesting me in faster uplink at home, it's offsite backups :-)
M.
>That's interesting because I am occasionally remoting-in to certain key devices using RDP over bog standard ADSL and the 1Mbps uplink speed of the originating end is absolutely a bottleneck with complex screens...
I suspect the difference is the headroom, yes I've noticed more dynamic things that your typical office application can take a little longer over the WAN, however, I suspect why I don't see things go quite as bad as you is because I'm effectively capped at 8mbps, so there is (some) capacity to handle the extra demands, with a 1~2mbps ADSL line, there isn't that headroom.
>though it isn't the RDP thing that's interesting me in faster uplink at home, it's offsite backups
Yes these are quite tricky...
For the remote home users (think aged parents etc.) I decided to invert things, they normally use a cloud drive - so for most things local 'cache' access, with the cloud drive handling the sync. of changes. I then implemented a backup of their 'in the cloud' cloud drive.
Obviously, this doesn't backup Windows et al. but if you've a basic/standard Windows+Office build, the recovery time from cloud backup isn't too dissimilar to doing a clean rebuild, particularly when (for these 'favoured customers') it is probably going to be a new machine...
For a business client, cloud backup (over VDSL) is meaning they are having to totally rethink their backup strategy.
For the remote home users (think aged parents etc.)
My plan is to have a NAS at ma & pa's, backing up their Mac simply using Time Machine, and one at ours which stores our media archive and network folders. Using OpenSuse I'm not that fussed about backing up the "OS" to be honest - all that's lost if the computer completely dies (or I decide to wipe it for other reasons) is local settings (home folders are still local), and sometimes it's good to have a clear-out.
I do have two NASes, but by the time I'd set up the backups and they'd completed over the LAN it was obvious I'd need to increase the storage in both pretty pronto and I've not (yet) got around to taking one to ma & pa's, "just in case" it's easier to rebuild it than expand it and re-synchronising 3TB over 1Mbit/s ADSL would take (I've just calculated it) about a year, assuming 80% of full-use is possible and no further data is added in that time :-)
Even just synchronising between the NASes when we've uploaded 50GB of holiday snaps and video would take about 5 days at 1Mbit/s - less than 6 hours at 20Mbit/s.
For a business client,
Had a similar situation at work a few years ago. We had a stash of (then) about ½TB of data (it's now considerably larger) that "central" IT doesn't want to know about and, in fact, ejected unceremoniously from their own systems without warning, fortunately just after we'd finished making a belt-and-braces backup onto external drive. We considered online solutions such as Amazon Glacier and the costs to store weren't dreadful, but we very quickly realised that getting at the backed-up data in the event of a total loss would be horribly slow and quite expensive. Instead we took that USB drive (and others, and local backups on local NASes) and "hid" them. One manager has it stored under his bed :-)
Since then, other departments with similar stashes of data that central IT were ignoring have managed to get together and a better, more accessible, more central solution is beginning to get off the ground (obviously with the help of IT) but it's been a very long, very hard slog.
M.
This is frankly PR to sell their kit which will make little difference to people on 100Mbps broadband and none to people on 20 Mbps or less Broadband.
Not at all. Go 15m from your WiFi router, and you'll see speeds drop below 20Mbps... Much lower with older tech. The newer kit will keep speeds higher as the reception conditions deteriorate.
At home, my laptop appears to connect to my wifi, but it doesn't actually work. I've tried using a repeater, but that makes my phone jump randomly between networks (even when one is sitting right next to it).
Even in a best-case scenario, there's a wide area that's close enough to the AP to make a phone try to use wifi and not use the cellular network, but far enough that the wifi doesn't actually work. If you happen to be standing there, you'll just be disconnected, and drain the phone battery.
A friend used to connect his first-floor desktop to the DSL router on the ground floor, but after changing the router it's no longer working. None of these issues ever produce any useful error messages; most of the time it just declares to be working, except that everything times out.
By comparison, my cellphone's 4G always works, if it's in a place that's covered.
https://xkcd.com/1865/
So, does this new AX standard actually, y'know, work?
The 4G where I live is very good and I tend to have the WiFi on my phone turned off.
This works fine until I get a system update that refuses to download until it can go over WiFi to save me from "unexpected additional costs".
I don't get an option to download over 4g anyway. This means that I have to park the phone close to the router.
I cannot get "unexpected additional costs" from downloading without WiFi. I have "unlimited data". I suppose it is not infinite and if I had high resolution Netflix running 24/7 I might get an email or two by the end of the month. I don't so a couple of GB is not going to make any difference. I can't be the only person with this. I am sure I had the option in the past but, as I was not unlimited then, I didn't do it
Now I want to download updates over the PhoneCo internet, it won't let me. Is this the fault of Google, phone manufacturer or telephone company?
I expect that if there is control over the wi fi features, then that would include forcing the router to use one particular radio frequency (numbered channel), but I don't know if you can do that on the client end as well - so that Device A will just not see Network B, which is the one that you don't want it to use anyway.
Or if the repeater is expected to give good service, then you could physically block radio signals from the router on a different floor. For instance, standing the router on the lid of a metal biscuit tin ought to stop it from transmitting downwards. Or you could put the client end in the tin, with the open end facing in the direction that you want it to be allowed to see.
A mesh network is a last-resort because it wastes an awful lot of time shuffling data around a slow, unreliable, shared medium. Adding "mesh" access points, particularly if you are in a crowded WiFi environment is just shouting louder at a party.
Think about where you need to use the WiFi. Run a cable to the nearest convenient point. Stick an Access Point there. Switch the one in your router off if it causes problems, or do as I did and buy a decent router without WiFi built-in. I haven't lived in a house in the last 50 years where the ideal place for the modem/router (as close as possible to the telephone master socket) is also the ideal place for a WiFi access point.
You absolutely can run several access points with the same SSID. You don't need special kit and your phone or laptop will connect to one or the other, and swap when it gets a better signal, but unless they are a good distance apart don't, for heaven's sake, run them on the same channel(s). Even if it means confining them to (for example on 2.4GHz) "g" speeds rather than "n", you will get a better experience from two non-interfering access points on a "slow" standard than from two overlapping ones on a "fast" standard.
"Auto" channel settings on ISP-supplied router/access points are the bane of my life around here. Every blasted house is using channels 1, 6 and 11 - often all three because of "mesh" or "repeater" products - when almost the entire globe other than the US would be better using 1, 5, 9 and 13. It makes it difficult to slot my own APs "in between" weaker neighbours. For example, if I had a strong neighbour on 1 and weaker neighbours on 6 and 11 I could look at using channel 7 or 8, which would only partly overlap with the weak neighbours) meaning that whereas in the early days I could cover the entire house adequately with a single AP, I now need two, and frequently channel 13 is the only relatively safe bet.
If I want to cover the entire house, that is.
And why is it that even the radios which can use 5GHz rarely use more than the first four channels? That makes 5GHz little better than 2.4GHz, particularly when some twit has it set to "80MHz ac mode" and blatts out the entire band. How hard can it be to implement the interference-avoidance that is mandated for the other eight channels?
Oh right. It's just easier and cheaper not to.
</rant>
M.
If you're going to run WiFi extenders as multiple access points don't forget that they need to be physically located *between* the main access point and you - half way is best. If the WiFi router is in the living room you're getting slow WiFi in the bedroom and you put an extender in the bedroom it will only get the same connection rate to the living room as you do. The reported rate to the bedroom extender will be faster, but the overall throughput isn't likely to go up much. There are WiFi units with claims about super-duper antennas and they might make a bit of difference, but locating it between would still be better.
Dunno - I have one of the new ASUS mesh WiFi products, with twin 802.11ax routers, and it works brilliantly. It genuinely has solved the coverage problems, and the whole house and garden now gets reliable coverage, without a hitch. I've measured the 'backhaul' mesh link at over 900Mbps, using netperf3, so the Gigabit Ethernet serving the wired devices actually becomes the limiting factor, which would have been improbably with any previous WiFi product.
Ok, well firstly you are using rather new and (relatively) expensive kit (ax) that very few people have at the moment, and a lot of "mesh" or "repeater" products are using older ac and n standards. Secondly I notice from the spec. that in order to achieve ~1Gbps link speed (which is effectively what you are claiming) it needs 160MHz of channel - at 2.4GHz that would take up the entire usable spectrum pretty much, and at 5Ghz it takes up a half of the total space available to properly DFS/TPC-equipped devices so I'd be interested to see what happens when all your neighbours decide to do the same.
Thirdly you imply that you don't actually have a mesh - you have what is effectively a single point-to-point wireless link (two devices). Using 160MHz channels you can get 1Gbps (ish) in one direction, but don't forget that as a shared medium you can't have two devices (actually) transmitting at the same time on the same frequency, so if you were doing simultaneous up- and down-loads I'd not be surprised to see them get about half that speed each, and adding further devices will reduce the rate further, though I note that ax is much more intelligent about sharing bandwidth and frequency re-use (where the radios and aerial configurations allow) than previous standards so is presumably much more efficient in typical "bursty" situations.
Wireless is (in my opinionated opinion) never the first answer for a connection problem. There are definitely circumstances where it is the best answer, but if it's possible to put a wire in, it's still usually a good idea to try that first :-)
M.
You need the hardware in your machine to support it, which probably won't appear in your work machine for at least a few years.
Most people I know have personal machines between 4-9 years old so no way they will support the standard without swapping out the wifi card.
Well, it may come as a USB module?
I struggled with the article, does "Wi Fi 6" on the hub improve performance if some or all of the client devices are built to a previous standard?
Of course, asking the question reveals that we have legacy devices at home. Like everybody else.
It may allow more clients transmit/receive at the same time - you may not see an increase in the top speed, but less issues when many devices are connected and communicating. Think about the difference between an hub and a switch (although it's not exactly the same thing, and there are more constraints on radio channels).
If you use many wifi devices at the same time you could see improved performances even with older devices.
But even if you are using modern machines, it doesn't mean you have a superduper WiFi card installed.
The big difference I've seen is that cheap home routers only supported 802.11g, which constrains many 4~9 year old devices with 802.11n adaptors. Upgrade that router to practically anything modern and suddenly that old laptop gains a new lease of life, because it can now fully use the capabilities of that 802.11n adaptor.
I came here to post something very similar: a length of Cat5e between your router and your desk will give much higher bandwidth.
And it leaves the aether free to connect those mobile/LAN-socket-free devices (alternatively, free to get cut off by the microwave oven you're using to reheat your lunch).
Not before time, as it's clear that most people doing mobile-phone or even laptop video calls for TV news, and so on, are relying on crappy wifi, given the choppiness and poor quality of most of what we're seeing. But at least it's encouraging to see that more people now seem to be getting the message that portrait video in a video-conferencing or for-TV context is for knobheads.
I doubt even a small percent of home workers are going to go out and buy a new router that supports WiFi 6 unless their employer pays for it. A large majority of people use the low end free router that their ISP provides, they plug it in and only ever think about it if their internet stops working and they need to reboot it.
Plus even if their employer was to pay for a new router there is a good chance they will also need to invest in new PC's and phone that support WiFi 6 if they are running devices 12 month or older.
Fortunately I had the house CAT 6 wired when I had major building work done about a decade back. Combined with a 72/20 Mb/s FTTC link from Zen we've been able to run two simultaneous video conference sessions with no problems (at our end). It's been noticeable that friends and colleagues with crap WiFi have real problems even at reduced video resolution.
I thought that, but using the new ASUS 802.11ax twin-pack mesh routers, it really is a massive performance leap. I've benchmarked the 'mesh' link (between Gigabit wired computers connected to each mesh node), and consistently recorded 940Mbps (using netperf3), so that's actually hitting the theoretical 'real world' maximum of GigE. If you use 'jumbo' frames you can get it a bit higher. So the claim of multi-gigabit 802.11ax mesh backhaul speeds might actually be true. Whatever it is, even with 'traditional' WiFi client devices, we're getting excellent coverage and performance all over the house and garden. I've been making some 12-way Jitsi Meet video calls without a hitch, from the other side of the house to the nearest mesh node. I mean, it's not Blenheim Palace, but still, the system really works dramatically better than our old single WiFi5 router.
> Combined with a 72/20 Mb/s FTTC link from Zen we've been able to run two simultaneous video conference sessions with no problems (at our end).
I should hope so, in our house, we are managing 4 people with Zoom/Teams sessions and concurrent iPad streaming etc. on an EE 36/9 FTTC.
Mind you in the home, with sub-20M runs, I don't see the benefit of Cat 6 over Cat 5e.
Mind you in the home, with sub-20M runs, I don't see the benefit of Cat 6 over Cat 5e
I thought that, but a friend offered me the end of a roll of Cat.6 he wasn't going to use, so I began to use that in our refurbishment before buying Cat.5e. I noticed two things:
1: the main benefit of 5e over 6 is that the former is cheaper (it's also easier to work with on a DIY-basis, but that's by-the-by), but many of the cheap 5e cables use CCA (Copper-clad Aluminium) rather than pure Copper, so if you are going to insist on Copper only, the price difference isn't as great as you might expect. I bought my own Cat.6 once my friend's reel was finished.
2: I ended up using a lot more cable than I was expecting. Partly this was down to asking the children to help pull and not supervising them (much), which meant that "leave a bit of slack at each end" seems to mean about 5 or 6m "spare" at each end of most of the runs. Partly it was simply down to the runs possible. It's not a huge house - "one and a half" storeys with the IT cupboard slap bang in the middle, but routing cables around all the other stuff in the place (the ventilation system takes up a lot of space and you wouldn't believe the amount of water and heating pipe we've installed) means that a run from the IT cupboard to the lounge - no more than 10m in a straight line - is close to 20m of cable from patch panel to wallplate.
Now I've no plans to use 10GbE, and I certainly can't afford the kit at the moment, but we're not intending to do any major work in the house for the next 30+ years so a bit of future-proofing seemed like a good idea.
M.
After I wrote my comment, I realised I was really referring the situation years back and an assessment of general performance demands.
From recent cabling work, installing Cat6 today is largely a no brainer, the volumes are there and prices are decent.
However, depending on ease of install, I would suggest if you have a supply of top-notch Cat5e drop cables (eg. from clients who upgraded to Cat6) that would otherwise go in the bin...
>Now I've no plans to use 10GbE
Given your total cable run lengths are going to be sub 30m I expect your installation to support speeds significantly faster than 10GbE..
Given your total cable run lengths are going to be sub 30m I expect your installation to support speeds significantly faster than 10GbE
Yeah, but even 10Gb kit is out of my (home) league at the moment and I'm struggling to think of a use case other than shortening the time it takes to transfer large video and photo files into and out of my NAS. Maybe in 5 years I could look at putting a 10Gb card in my NAS and another in my main computer and getting a switch with some 10GbE ports but at the moment that looks like a £1,000+ option so a second 1Gb card in the NAS and a bonded/trunked connection would ease the situation for nowhere near as much money - thirty quid for an Intel-based NIC and my cheap switch is already capable of (limited) port trunking :-)
That said, and I realise futurology only works if you're Harry Seldon, but in a domestic situation, even in 10 years' time I suspect that if 10Gb is widely adopted, it'll mainly be because manufacturers are just fitting the parts as standard, rather than for any practical reason.
I was going to write that 10 years is a long time in computing and that amazing things happen in quicker timescales, but I'm not so sure that's true any more. I bought my first stand-alone 100Mbit switch (i.e. as opposed to the one in the back of an ADSL modem) some 16 years ago and felt bad because a friend of mine had "saved money" by buying a 10Mbit hub. He saved about £15 over the cost of my switch. I bought my first 1Gbit switch perhaps 10 years ago for about the same price. 10Gbit switches - even those with just a few 10Gb ports - are still ten times that cost or more. By rights it should be possible to buy a 5 or 8 port desktop 10Gb switch for less than £50 by now.
Maybe it's the lack of a use case?
On a different subject, my last "main computer", based around an AMD A10, lasted seven years before I replaced it. It would have managed another one or two if it hadn't failed (processor died, probably because of dust-related stress) and the new machine keeps all the components other than the processor, motherboard & memory, and an indulgent NVMe SSD.
M.
It might, but I wouldn't count on it. The larger companies should have the IT chops to allow people to work from home, but will they have to will to do it is the question. I'm thinking not, because having the employees on-site is generally better from all points of view, not to mention security. Having all those people working remotely must be giving the heebee-jeebees to IT admins everywhere.
Then, of course, you have all the companies that simply can't work from home : plumbers, construction workers, supermarket employees (and stores/restaurants/bars in general), medical companies, etc. Basically, if your job is to deal with the public, you're not working from home.
Finally, you have all the consultants in various capacities. They can't work remotely because the clients want to see them there. Sure, right now I have customers allowing me to work remotely, but that's because they don't have the choice - there's work to be done and it can't wait. When the confinement is over, I've already been told in no uncertain terms that I will lose my access. Security first.
So no, I don't see that there is going to be a change in the way we work, globally speaking. Yes, some companies will realize that they can reduce their office space requirements by having people work from home but, little by little, the manager paranoia is going to creep in and the question of are they actually working is going to haunt them. And that will be the end of the experiment.
I've worked from home since 2005. I generally go into the Office at most twice a month. my customers don't need to see me in the office as they are all over the world. Japan, Brazil, Middle East, India, USA etc.
My employer likes people to work from home. Saves them lots of mostly unused office space. My manager is in Atlanta so we video conferece every two days. I have all the kit I need to do my work. This current situation has really made no change to the way I work.
Naturally, every company is different but I really would not want to have to commute again. I did the spam can trip into Victoria for twenty years. Life if a lot less stressfull since I stopped the slog into the office every day.
As he pointed out.
Do you work in an office type environment. If so there is a chance you can work from home.
If not, you probably can't.
I said id work from home, just a tad hard to work on F1, Hypercar and supercar engine blocks in my lounge though.
"Basically, if your job is to deal with the public, you're not working from home."
Broadly true, but the significant exception is call centre workers. A couple of our clients are finding ways to allow their call centre staff to work from home via VoIP over a VPN.
I don't think we've worked out a way to make it PCI compliant yet, so they'll still have to have a skeleton staff actually in the office.
As to whether this will continue past lockdown, for our customers I think they'll probably prefer to have people in the office. However, the call centre industry in general seems to be pretty cut-throat, so if there's money to be saved from having your workforce at home I bet they'll try it.
We already have partial WiFi6, via one of the higher-end ASUS mesh router products. It works very well, even though for now, the clients (i.e. phones, etc) are connected using traditional WiFi5/4, since none supports 802.11ax yet. But I've measured the WiFi6 mesh connection speed, between the twin boxes, and repeatedly recorded an impressive 940Mbps using netperf3 (with the boxes around 8m apart, in neighbouring rooms). This was between two hard-wired computers, one plugged into each node. This actually suggests that the GigE connections (which top out around that level, minus the overhead) are the limiting factor, and the mesh link really could be achieving gigabit-plus speeds (it claims up to 4.8Gbps on the box).
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
