Hacker News new | past | comments | ask | show | jobs | submit login
Wi-Fi 6 is designed to reduce congestion from devices (nytimes.com)
128 points by CapitalistCartr on Feb 25, 2021 | hide | past | favorite | 185 comments



Ok, a few things.

1. WiFi 6 only improves / reduce congestions when all devices on the network are WiFi 6 only. As soon as you have older WiFi devices, the difference becomes negligible to non-existence.

2. WiFi 6 only improves / reduce congestions when all devices on the network support WiFi 6 OFDMA, which while being officially part of WiFi 6, it was not mandatory. And if you have one that doesn't support it or does not have a firmware updated to support it, read point 1.

3. WiFi 6E will mandate all those optional features that were intentionally missed out on WiFi 6 due to all sort of technical, economical, political reasons. So in realty if you dont have all your devices as WiFi 6E, read point 1.

4. It is not clear whether 6Ghz support is mandated to be certified as WiFi 6E. On paper it seems to be the case, In practice most part of the world dont have 6Ghz spectrum ready. And I am not sure how FCC ( or similar ) clearance will work for a product that are already shipped on the market. Could we get new spectrum support via Firmware update? I am sure that is how US intends to deal with it since Intel are already shipping WiFi 6E product with label that support 6Ghz. I am just not sure if EU, UK, or other part of the world would allow or follow similar route. That has an implications on how fast WiFI 6E could launch worldwide.

5. Finally, there is only so much you can do with WiFi spec. If you have an extremely noisy environment nothing could really help. In an ultra packed City like HK, I could detect at least a dozen AP at any given time.


> In an ultra packed City like HK, I could detect at least a dozen AP at any given time.

On a rooftop in Philadelphia, just standing in one spot and turning 360 degrees in a circle with my laptop, I was able to detect over 50 Wifi APs... (Good 'ol Netstumbler.)

This was in 2003.

A dozen in a city like HK seems quite low.


> A dozen in a city like HK seems quite low.

...which is the actual effect in a high interference scenario. The more interference you have, the less interpretable data you get.


Also you don't know how many people turned off their WiFi and just ran Ethernet cables everywhere after struggling to get usable connection speeds.


I'm also in HK. Keep in mind that there are solid walls and floorings between my home and other apartments and buildings.

I can count 67 APs right now.


Well I didn't double count all the 2.4/5Ghz Variant. But even so, 67 is.... insane.

And Oh Wow. I am checking out T1 now. :)


Japan in 2007 on a fairly high floor the list went off the end of my mac's screen. And my bluetooth mouse did not function.

2007.


Your comment resonates with me. From our rural house in Ontario there are no other APs visible. None, but our own. I've even started disabling encryption just because I can. But we struggle to get even basic internet connectivity. Interesting that on both ends of the spectrum there are issues.


Starlink is not available in your area?


starlink only just went publicly available - its going to take a while to filter out to people who it could benefit.


Just got a Starlink invitation. But it has very expensive monthly rates, that are more than twice the rate of standard isp's. In rural areas. $136 in Canada. Plus a lot of hardware. (625$) It is very cool, though. Haven't decided. At this time - Only worth it as a tech geek out, experiment. (Unless it gets better and cheaper.)


I see 24 APs other than my own just sitting here on my couch in a locale less dense than HK. So yes, agreed


>> It is not clear whether 6Ghz support is mandated to be certified as WiFi 6E.

It is.

>> Could we get new spectrum support via Firmware update?

Yes. In US/FCC, the procedure is called "Class II permissive change".

Also, US, Korea, EU/UK, Chile, Brazil have opened up 6 GHz band. More countries will follow in coming years.


I am confused. I thought "permissive change" referred to the ability of a device maker to avoid having to do a full fcc recertification when making changes to their device (hardware or firmware) that did not affect rf emissions


My mistake here actually. They still need to go through FCC testing but they can enable already deployed devices through SW/FW updates.


Correct me if I am wrong it was a long time since I checked the progress.

AFAIK, EU/UK, Korea only has plans to discuss and move forward on selected spectrum sometime in mid 2021. No exact date has been announced unlike the US. Which is a done deal. i.e It is not likely any part of the world other than US to get 6Ghz in 2021.

> Class II Permissive Change

Is that new? I remember Smartphone cant add new frequency spectrum support without going through FCC recertification again . Although that was many years ago.


Korea has rules in place since October, 2020. Certification has started. Samsung Galaxy S21 supports Wi-Fi 6E in Korea now [1]

UK/EU is in final stages of regulation. Final approval for EU countries is in March 2021; and after April, countries will implement it in their rules.

Brazil has approved their rules today! Chile and some other countries have rules in place as well.

[1] https://www.fiercewireless.com/wireless/samsung-galaxy-s21-u...


Wifi 6 feels like the transition from 802.11b to 11g all over again i.e. having a single b-only device will cause your entire network to fall back to 11Mbps for compatibility. At least back then those devices were either quickly replaced or at least had upgradable wifi cards, and we many more monolithic phones and IoT devices to deal with.


802.11b to 802.11g was not much of a transition, just higher speeds without any real protocol improvements. Back then when I was using long distance (~4 km rural wireless hops -- was too far for DSL) 802.11g links, they were atrociously unreliable. It wasn't until 802.11n gear finally came out is when wireless finally began to work reasonably well for me. MIMO meant that reflections and other oddities were better compensated for. Moore's law strikes again - more transistors, more math, better performance!


> 1. WiFi 6 only improves / reduce congestions when all devices on the network are WiFi 6 only. As soon as you have older WiFi devices, the difference becomes negligible to non-existence.

I assume that can be worked around by creating a dedicated network for WiFi 6 devices?


Yeah, this is a super interesting question. Does anyone know if this is the case?


Figure one thing about the 6ghz band is it'll be new and less congested.. at least for at least a little while. Right now i use one of those DFS "radar ones" which seem unused by most nearby APs and i see a lot - i assume it's manually opt in/more expensive equipment.

I'm on the 24th floor in downtown vancouver and see 45 APs and none are using the DFH bands.


Uk has already approved 6ghz


I have a Ubiquiti Unifi UAP‑AC‑HD wifi access point for my home. It's not WiFi 6, but it can handle 500+ concurrent users while delivering 800 Mbps on 2.4 ghz and 1733 Mbps on 5 ghz.

All in all, I have around 70 wifi devices in my apartment. That might sound like a lot of devices (it is), but I'm a nerd and I like to tinker with lots of smart home stuff and I also have a lot of computers and gadgets.

Also, because I live in an apartment complex, there is a ton of WiFi noise from other neighbors. There are around 20 different WiFi networks within range.

I went through multiple WiFi routers (including high end $300 ones), but I was rarely able to get over 100 Mbps.

When I upgraded the Unifi UAP‑AC‑HD, it was like flipping a light switch. All of my wifi problems went away, and I was able to get 300-400 Mbps in pretty much every room.

I get that WiFi 6 is great and all, but at this point, I'm not really sure what I would be gaining by upgrading. The UAP‑AC‑HD is on the high end of the previous generation, and at least for now, it's good enough for me.


> but it can handle 500+ concurrent users while delivering 800 Mbps on 2.4 ghz and 1733 Mbps on 5 ghz.

i feel you, its much better than your previous ap's but please dont buy too much into the marketing.

(single dualband-ap can at best handle >200 concurrent devices, but not active users)


And those throughput figures will be shared among all the users. Not available to each and every user, as it seems to imply.


I don't think anyone in their right mind interprets that as 800mb/s or 1733mb/s to 500 users simultaneously. It's pretty clear what it means.


What may be pretty clear to a majority of HN readers may not be so clear to the vast majority of consumers


> please dont buy too much into the marketing.

I'm confused, I thought the entire premise of the comment was that they actually measured the performances of these routers instead of looking at marketing?


Nah, simple fact OP cited 802.11ac marketing wank numbers means nothing was measured. For starters those are PHY Speeds, bits in the medium (air), that doesnt take into account overhead. AC has 65% _best case scenario_ efficiency (only two devices practically sitting on top of each other), while N had 80%. Then you get up to 10dBm worse SNR at top speeds compared to N.


>> I have around 70 wifi devices in my apartment. >> there is a ton of WiFi noise from other neighbors.

Is there? I wonder if your neighbors are having similar problems?


Generally wifi noise "from neighbers" Comes from a bunch of independent radio groups that happen to use all of the specture. When you have a single AP home it doesn't matter if there are 5,000,000 devices on it to your neighbors it's when you get that second AP in another channel that's noise.


The reality is that these manufacturers aren't making their own WiFi chips. They all use the same few WiFi chips from the same few vendors.

There is room for variation in how the products are implemented, but there's no real reason why a consumer router shouldn't be delivering the same high rates unless there was something else weird going on.


A lot of WiFi hardware implements a substantial portion of the stack in firmware or driver blobs.

So even "the same few chips" can be manipulated to perform very differently. Stuff like retransmit, frame timing, power/gain control, and even tx/rxchain selection and MIMO configuration are often done in software and can be altered between suppliers. Additionally, vendor drivers often perform very poorly at the higher levels as well - the MAC level (associate/beacon/disassociate), the framing level, scheduling encryption/decryption, keeping track of associated clients, and so on, are also all often handled in software and can be optimized.

Plus, once we reach the router level, we also get into the often abysmal IP stack configuration on consumer-grade routers.

In short, there's a lot that can differ, both in IP-land and in chipset-land. Anyone who's "fixed" an ailing consumer router firmware by reflashing it with something else can understand the IP-land pieces, and a quick read through a driver written on the Linux SoftMAC stack (for example) can really illustrate how much goes on at the driver layer (not to mention firmware).


Vendors provide reference software for their chipsets. More often than not, the software provided by vendor is the bare minimum needed to prove that their hardware sort-of-kind-of works. It is by no means able to achieve the best performance out of a given system.

Source: worked at Ubiquiti (not on networking hardware though).


as someone who did windows device drivers a decade ago this so much - i took a example DDK driver and rewrote it entirely and the resulting performance gain was very noticeable. Not to mention at the time on the device side working around hardware bugs to take something from "yea it works" to "it works really really well!"


It's true they use the same chips, but board layout and connectors become important as you get to higher frequencies. Wifi antennas can be anything from PCB stripline to external whips.


That’s not true. Antenna design for these 5 ghz+ frequencies isn’t trivial at all.


My parents have a Ubiquiti as well. Don’t replace it until your friends have faster WiFi. It’s one of the best units I’ve ever used in a home. Probably the best, actually.


> it can handle 500+ concurrent users

no, it can't, try putting 60 actual laptops moving a moderate amount of traffic on it... ubiquiti and marketing hype have a near 1:1 venn diagram overlap.


I used to manage the network at a co-working space and I had two of these access points installed in a mesh setup, with one covering the main front room, and one covering the back part of the building.

I'm not sure if it can handle 500+ concurrent users, but we would frequently have as many as 100 people present and on the WiFi during some of the events we held, and we never had any issues. We would have people live streaming video while other people were working on their laptops. As the network admin, I regularly did speed tests and asked people how the internet was working for them and everyone was always very happy with it.

Over the course of about two years, we had over 3,000 people log into the WiFi, and never once had a single problem (other than from our main gigabit cable internet going down once or twice).

This is a HUGE contrast from my experience with consumer grade WiFi routers like Linksys, Netgear, and D-Link, which often advertise crazy speeds like 2,000 Mbps but only deliver 50 Mbps in real life. I once had a $300 D-Link router that I literally had to reboot every week or else it would become completely unusable and slow to a crawl, for example.


Ubiquiti is still pretty good for home usage. Their gear is the only one I have used so far that handles 10-15 active devices without breaking a sweat. Other APs I've tried fall apart pretty quickly at that number.


Any clue "How" did it fix everything?

With everybody's router blasting at High Power, what can you actually do on your side with just one router to compensate?

Thx muchly!


Nearby networks won't harm your signal to noise ratio unless there is active traffic. Most of the time, 90% of the 30 SSIDs you see nearby are mostly inactive.

Anyway, Ubiquity and commercial gear in general just has better software and hardware design.

* The UAP‑AC‑HD has been around for at least 5 years and still gets updates. Wifi is full of weird bugs so its impossible to release a new chip set bug-free. Consumer gear just doesn't get the same duration of support as commercial. * These access points are designed specifically for setting up networks with multiple APs and lots of nearby networks. * These APs are just access points and offload routing and even their UI to a router and management server. A lot of consumer gear uses/wastes a lot of RAM on their fancy management interfaces. * Exterior profile stays the same. Consumer gear companies spend a lot of time & money on marketing and redesigning the 'look' of their products which diverts focus. * Better management/software: All the Unifi APs use the same management server software which consolidates features and fixes. The software also has a lot of performance related features like disabling/limiting the slowest speed enabled on the 2.4Ghz and 5Ghz network.

Really the last point is a really big one; consumer gear (besides new consumer mesh networks) is often sold on having great range/coverage but these can cause the whole network to be bogged down by a single device operating in 802.11g or older mode. Commercial gear expects you to use multiple APs to satisfy coverage but has the features to prevent network degradation.


It's really not power, it's just debugging and optimizing rates and retries. Lots of wifi has lots of dumb bugs that no one ever figures out, because stuff still mostly works, with all the built-in retries.

(I used to work for an enterprise WiFi AP vendor, debugged some driver bits myself ... back when 802.11n was brand new, all the enterprise vendors were increasing their effective throughput from 80mbps, to 100mbps, to 140mbps, etc, just by debugging and optimizing their drivers.)


Use DFS channels. I have a friend who is a streamer and lives in an urban setting... almost 100 5GHz APs are visible in a scan in her apartment. She was having constant issues that we tracked down to huge latency spikes due to the congestion for the non-DFS channels. It turns out that her cable company's wifi AP / cable-modems do not support DFS.

I got her a wifi AP that supports DFS, and she has not had any issues..

Note that I initially got her a fancy Netgear which claimed to support DFS, but we could never get it to work. I spent a week going in circles with their support and eventually returned it.

I eventually just gave her my several year old Atheros-based TP-Link router running DD-WRT. It was able to use DFS channels, and she has a channel to herself now.


Along with what everyone else has said some vendors such as Meraki and Ruckus have separate radios just designed for monitoring the spectrum and making changes accordingly. Groups of APs will automatically decrease their power and channel swap to avoid channel overlap etc.


Ubiquiti is enterprise grade gear, so first thing is this is not a router, it's just an access point, and it's optimized for lots of devices, lots of interference typical of office scenarios.


An an owner of a decent amount of Ubiquiti gear, it’s “enterprise” in the sense that the management system scales to lots of devices, and that’s about it. Every recent revision of the management tools makes it shinier and less usable.


For enthusiasts and small and medium businesses it is fine, but it definitely is not enterprise grade.


Agreed with the other comments here. I couldn't stand the Ubiquiti mgmt interface for home as it needed a VM or chroot plus a database instance, etc.. When I switched to an equally spec'd Ruckus, I was blown away at the performance difference. Also the mgmt interface a simple web interface. I couldn't be happier with my switch and will probably be sticking to Ruckus for my home for the forseeable future.


and at my initial google 10x the price lol


It’s pro-sumer, definitely not enterprise grade gear at all.


The Ubiquiti stuff has a lot of automatic channel selection stuff built in to use the best frequencies for your conditions.


Most APs do this already.


Most APs do this once at startup.

Gives you good benchmarks when you start it but over time fails in a noisy environment.


So cell breathing is effectively prevented?


I guess it comes down to how well they do it.


Maybe, but my experience with consumer grade network gear's software has left a lot to be desired. I'll never bother with Netgear/Linksys crap again.


> I was able to get 300-400 Mbps in pretty much every room

I thought it was 1733 Mbps on 5 ghz? Is that figure the throughput sum over all concurrent devices? If so, could a single device with multiple antennas get 1733 Mbps rather than 300-400?

Personally I've never been able to go much above 400Mbps on Wifi 5 with any device. I'm hoping Wifi 6 will change that.


1733 is a bullshit marketing figure. Nobody has ever seen that.

The best I’ve ever seen on WiFi 5 is 850-900 MBps (4x4 MU-MIMO router, 3x3 MacBook Pro, nothing else active on the network)


I’ve bought one of those to upgrade a Mac airport. It’s rubbish! I get much faster real world rates on the old airport and it doesn’t fall off with range so quickly! Very disappointed. Great marketing though!


I see that the Wifi 5 access point that you write is more expensive than their Wifi 6 offering, and it's advised for outdoor scenarios.

Which is the best router for long range and just few decices?


Any router with a directional antenna pointed in the direction you need.


How far is long range to you?

You might be better off to do a Station->AP wireless shot with a couple of Nanobeams if you're getting crap signal, if there is electricity where your signal is crap. If you'd like to use Ubiquiti, that is.


just about 15m, but it needs to go through a wall. It’s not even a big apartment, and the walls are not that thick, but as it’s in 1 direction, I’ll try to get a directional antenna as the other commenter advised. Also probably any long-range device will be better than the cheap Wifi 6 Archer that I bought (although it was a great improvement for short range ping time compared to the Wifi 5 router that I had before)


If you manage get the signal from the AP to the device, just remember the device has to be able to talk back.

If the device is low powered, this may still be a problem.

A multi AP system allows the higher powered APs to talk to each other, and the lower powered devices to talk to the nearest AP.


sure, I’ll try that, but when I tried it before, the added latency was as bad as the original connection (though the repeaters didn’t have external antenna).

Quite often the biggest problem is not the transmission speed / latency, but my Samsung Note 10 reconnects to the AP randomly.


If it’s headed through a wall, you might be better off putting an AP in the room that it will be used and backhauling over power line adapters or a MoCA adapter if you have some unused coax drops. It doesn’t seem like you would get good performance out of power line adapters in that application, but my experience has been pretty good. You can even find Homeplug devices that also do PoE, so it’s minimal wiring at the AP.


What is it? How is it different from wifi?


A wireless access point is just a more accurate accurate term for a wireless router.

If you're asking what the Unifi UAP‑AC‑HD is, it's a relatively inexpensive, prosumer, wireless access point from Ubiquiti. The main appeal is: the Unifi ecosystem (optional, but it does offer extra capabilities for your network) and PoE connectivity (ethernet cable is used for data and power). I have two Unifi AC Lites (a slightly less capable model than the HD) in my home they work quite well (and also look good).

On the professional end, Ubiquiti provides wifi hardware for commercial settings like stadiums and conference halls, so they're well versed in creating hardware/software that deals with hundreds or thousands of users/clients. They also offer long-distance communication equipment (think 'wifi' over hundreds of metres)


To be more precise most serious business grade wireless access points such as the Unifi UAP series are not actually routers, they're layer 2 ethernet bridges onto a single switch fabric, or a varied set of VLANs in the case of an AP that broadcasts multiple SSIDs and connects users to different network segments.

The routing is done elsewhere, either at the switch that the AP is plugged into, or something another hop or two upstream of that.

The typical $50-100 "router" with built in 802.11ac wifi that you might buy at best buy is indeed a router, with both wifi, some basic routing/nat ability, and a dedicated WAN interface.


The Ubiquiti hardware is fantastic!


to be honest, any moderm AP will do the job. processing power increased by a lot recently and the spec is mostly just narrowing down what enterprise gear used to do for years as an "extra". e.g. wifi6 (ax) is just officializing beam forming that every premium ap had since wifi-n but was not part of the spec.

praising unifi you are just suffering from new-shiny-toy-syndrome. Specially because your unifi AP is the only (few?) product on the market that requires a second appliance to work (and don't even get me started on running the controller on a vm with outdated dependencies, out of sync open source code, and broken mongoDB implementation) to not have more than half of the advertised features disabled (e.g. vlans, guest, etc)


> praising unifi you are just suffering from new-shiny-toy-syndrome

I was responsible for a small office, and I tried a bunch of different big-brand expensive prosumer APs, and they all had troubles.

I then tried Unifi gear, which wasn’t as easy to set up, but it worked flawlessly and has done so for years. I’ve since used a few different Ubiquiti products in a few other situations and the gear has just run solidly with zero trouble.

I have used other reliable gear: the Apple Airport was great but now discontinued, every Fritzbox I have dealt with has been fine, and I currently have a Mikrotic at home which has been reliable (albeit in a very undemanding environment). I am about to add a outdoors Ubiquiti AP at home.

Maybe gear is more reliable now, but I recommend Ubiquiti to friends and it hasn’t let me down yet.


Mikrotiks are good, and they are cheap, but they are very hands-on in setup, you have to know what are you doing.


I have a pretty good idea of what I'm doing, but still find Mikotik confusing and hard to set up for non-trivial use cases. Like configuring an SSID per VLAN.


Which AP are you looking at for the outdoors ?


Unifi APs being bridges only is a feature, not a bug. There's a wide range of all-in-one devices on the market; pure AP bridges, not so much.

And before you say, that you can configure all-in-one to work in bridge mode: yes, some of them. Many of them cannot, or have weird limitations when you do.


> Unifi APs being bridges only is a feature, not a bug

I think OP means that you have to be running Unifi somewhere in order to get all the features from the AP. Not that it's just an access point and requires a separate router (which I agree, is a huge plus).


Strictly speaking you don't need to run the controller. The Android app can configure APs in standalone mode and update their firmware.


He wrote "some features", not to run the APs in general.

You don't need controller per se to get per-user VLAN assignment or WPA Enterprise. What you need is running RADIUS server somewhere, and controller conveniently provides a basic one (it has FreeRadius underneath, but provides only basic configuration options, so you won't be able to use AD for your users; also do not edit the config manually, any update will wipe your edits).

For guest portal, you need, well, guest portal somewhere. The guest portal is a part of the controller. However, if you need guest portal, I don't see any problem running controller somewhere.

For the rest, you can also run controller on your computer once, configure everything and then shutdown the controller. All the devices will work fine without it.

Edit: word


you are correct. can't expect unifi fanboys to pass a reading comprehension test.

for the record, most other industry that offer APs that run on standalone will work with all features, or only advertise that feature on the package. For example, you don't see cisco(?) unleashed mode APs advertising guest portal, unless they can do it on unleashed (stand alone) mode.

Unifi is intentionally misleading.


Don't worry about that; even the biggest unifi fanboy will turn back into normalcy once they will have to contact the ubiquiti support (yeah, ubiquiti support, that's oxymoron).


I don't know about that. I use the second generation nest wifi and its struggling with the ~35 devices in my home.


How many of the mesh nodes do you have? I have three including the primary one, and have had as many as 50 devices on my network and it has worked flawlessly. Many are low bandwidth IoT devices/hubs, but I've literally never had any problems over three years.


About 30 of the devices are low bandwidth IOT devices, I have one hub and one node.


well, google consumer products are not exactly know for being "premium" in anything.


Great, but my local ISP monopoly will be instituting data caps in 2022 (only after massive public pressure to delay) and therefore speed won't be my major issue but instead overall data consumption. In fact, I'm looking into routers (UniFi Dream Machine) to throttle devices in order to prevent them from streaming 1080/4K video.

I don't need technology to increase the speed at which I exhaust my data cap, I need legislation to outlaw data caps and create ISP competition.


You should also keep an eye on idle data usage. I had several chromecasts plugged in around the house and discovered (with my UniFi setup) that they were each using about 500MB per day each just sitting idle. That's about 15GB per chromecast per month. Not trivial when you have several of them and a 1TB cap.

More details: https://www.howtogeek.com/337719/how-to-tame-your-chromecast...


> 500MB

What the hell is a idle chromecast doing everyday? That's absurd. Is it outbound WAN traffic or just local network traffic.


probably the images it downloads for the backgrounds it shows while idle? 500mb is a lot even for that thou


is that a lot? That's only 20 images of 25MB, and given that these images are supposed to look good on a large TV screen that large size seems reasonable.


Agreed, unfortunately my router does not give me this data, so I have to go device-by-device. I'm hoping the UniFi Dream Machine will solve this both by allowing me to throttle, and giving me the data to address individually.

I did install a firewall/data logging app on a spare Amazon Fire HD8 and found it downloaded several hundred megabytes of advertisements over a week's time despite sitting in a drawer unused. That was an easy fix.

Another glaring data hog was an Amazon FireTV stick which was set to the highest quality.

Overall it's been an eye opening experience to see how many devices consume bandwidth with no regard for optimization or giving users control.


i have a UDM-pro and it's done a decent job at that - it's really neat tbh.

But only get a UDM pro if you don't need link agger or the ability to control multiple networks from the same controller (i wanted to throw a USG at my parents but can't join it to the UDM pro because.. reasons). it has some annoying limitations (IDS/IPS cannot keep up with my internet anymore) that if i knew about i'd have gotten a USG-pro and a cloudkey.


Wi-Fi for home use was pretty well solved with AC nearly a decade ago now. The newer solutions focus on high desisty busness (local data need or call center/auditorium density) and newer direct high bandwidth device<->device connectivity.


> Wi-Fi for home use was pretty well solved with AC nearly a decade ago now

This is only true for small houses with certain construction materials, few devices, and an internet connection under, say, 50Mbps. The last year has been especially good for seeing how rarely that's true for most people, however. It's so easy to find people talking about how upgrading from a 10-year old access point or adding 1-2 more made a huge difference for things like call latency or contention between devices. Having everyone home and trying to work, attend class, or socialize online made dealing with all of that a priority.


Aren't you saying pretty much the same thing? Upgrading from a 10 year old AP almost certainly means upgrading to 802.11ac since Wifi6 AP's are so rare.

I don't know when 802.11ac AP's became common, but the first iPhone to support it was the iPhone 6 in 2014.


Not really – note the “nearly a decade ago” in the comment I replied to? AC was a win, yes, but it's not like the entire industry stagnated for a decade. Upgrading from even a 5 year old access point to a new one — or especially the scenario I described going from one to multiple in a mesh or backhaul setup – will make a big difference for call latency and gaming, simultaneous use (especially if you have a bunch of old IoT devices like my Toshiba orphaned TV), etc.


Apartment buildings can have very noisy public spectrum which 6 will help with


Unless people are sharing access points, WIFI 6 doesn't really help with this use case. You wouldn't get to take advantage of OFDMA or MU-MIMO in this scenario. At that point it's basically 802.11ac with 1024-QAM.


i'd say the average house has only one wifi-ap and since 5ghz (AC) just barely pentrates a solid built wall, 2,4ghz-N is where the majority of wifi usefullness originates, which is even older ...

but yea, cell confinement and directional interference optimisation is the way to go for high density applications.


In terms of frequencies used for WiFi and the dates standards were ratified, 5GHz is technically older. 802.11a was ratified just before 802.11b, but came to market a little bit later and hardware was a bit more expensive.


> 802.11a was ratified just before 802.11b

true and nobody used it, hence i was talking about AC which was practically the start of 5ghz wifi


I was strongly looking at a UDM this week. After reading through many testimonials: everything is not as it seems in unifi land. I went with an edgerouter-4 and will continue to use my R6700v2 as an AP. It’s a fine AP but a godawful router. It took disabling default configs to hit gig throughput. Forget enabling QoS. The WAN DHCP client also shits itself once a month. Enough is enough. The networking kit should be a box that sits in the corner and does its job.


> everything is not as it seems in unifi land

Go on... A buddy of mine just dropped several G's on UniFi gear so I'm hoping he can tell me about any issues before I jump in.


A bunch of my co-workers jumped on board, and with a good discount I gave it a shot as well.

When it's working fine it seems marginally better for wifi connections than a standard cheap consumer Soho router.

But that's the problem: it very often just doesn't work correctly. Port forwarding rules will sometimes, randomly, no longer apply. DNS issues will mysteriously crop up. Some clients seem to get a cut in downstream speed but not others (with or without enabling the setting to optimize channels). UPnP, if you choose to enable it, will mysteriously work for some clients but not others. Sometimes these issues will only happen for wireless clients, sometimes for wired, and sometimes for only a given port or AP.

It often feels like a double-NAT situation, but somehow with a regular, basic, client-only topology. My co-workers have reported both these issues and ones I have never seen.

The worst part is that you will go on the forums, sometimes find users with the same issue as you (and many responses) and then rarely see an official response beyond 'yeah that's a known bug, no eta'.

I've heard nothing but praises from friends for their other hardware. It's just the UDM that seems strangely awful.


The UDM line is about a year old and many features are in the "to-be-implemented" bin. This is a vague statement and I don't have specific examples on hand. I just recall seeing references to them when hunting around. I don't have firsthand experience.

The UDM can't be adopted by another controller: it needs to be the unifi controller. This really kills the upgrade path imo.

One of my best metrics for deciding what hardware to get is to look through the 1-star reviews for realistic testimonials by reasonable users. It takes a lot of sifting, but it builds an okay picture. There are almost no complaints about the ER-4 compared to the UDM. I think I'd likely roll a pfSense box before jumping into unifi.

https://www.amazon.com/product-reviews/B081QNJFPV/ref=cm_cr_...


I got a UDM-pro and overall while i am very happy with the other unfi gear i have i would have not gotten the UDM pro if i'd known about the limitations before hand: can't adopt USGs to its controller, link agger doesn't work ("yet" so never), and IDS/IPS maxes out at around 800mbps and i have gbit now so its limiting my internet.

i'd have gone with a USG-pro and cloudkey 2, but otherwise the unfi ecosystem has been good, i like the single plane view of my network and works well enough to not want to give that up


Note that EdgeRouter is not a product in their UniFi line, the most commonly praised.


By 2022 they may feel enough of the competition pressure from Starlink to abandon that idea. Hopefully yours and every other ISP out there.


If starlink can maintain urban bandwidth, more power too them. But I think that 1/2/10,000 dishes per square mile will probably overwhelm the handful of sats flying over a town at any given moment. Starlink is no threat to wired services in urban areas.


If the local regional fiber ISP gets their act together that will motivate them to run fiber to our region. Otherwise I'm not holding my breath. I may have to supplement my wired network with a low cost cellular based network.


Maybe someone more versed in UniFi will step in but from what I've experienced throttling in UniFi exists across the entire network, not just WAN egress traffic. It severely limits it's usefulness.


Pretty sure from what I've read it allows per-device throttling. I'll try to dig up a link.

Update:

Throttling via groups: https://help.ui.com/hc/en-us/articles/204911354-UniFi-How-to...

While this isn't exactly "per device" throttling it looks like it can be used to achieve the same thing. I would most likely have a "low speed" tier that I assign to things like smart speakers, kids devices etc.

I swear I saw some screenshots or maybe videos with a per-device throttle option but maybe I'm getting my routers confused.


What's your local ISP's name?


Probably starts with an X and ends with a finiti.


Yes. The only other option is a regional fiber network which requires thousands (if you're lucky) up front to dig a trench to your house not to mention a very high monthly fee (which I'd gladly pay).

I routinely use between 1.4-1.8 TB of data each month. I have about 30 devices connected to the network at any one time, maybe 50 total.


Do they offer a “business” connection? Often costs more for the same cable but is unmetered and comes with a static IP.


So, NYT dude is not very much impressed, but at the end of the article he says he is seeing dramatic improvements in latency of his light bulbs and garage door, which apparently took seconds with his old Wifi.

That sounds odd.


I've always felt that's why this kind of reporting is pointless. I don't care what some guy I don't know, who doesn't post his credentials, says about a technology that's very complex. I'd rather wait for Wendell Wilson to do a video and I'll watch it on Level1Techs channel on YouTube.

This whole model seems pointless and strange to me. We no longer need "tech journalists" when you can have literal tech experts creating videos that cover all this stuff in far greater depth.

I'd love to see the metrics on this article versus the best tech expert YouTubers.


Most people have no idea how to evaluate YouTuber credentials, and they subscribe to outlets like the NYT to do the filtering for them.

You're not the intended audience for these pieces so it's not weird that you don't find them very useful. Most people just aren't interested in the depth of detail that you are.

Imagine if they replaced all the articles on topics you are just peripherally interested in with wonky deep-dives. You'd probably give up after the first few because you aren't equally interested in everything, and sometimes you just want a breezy high-level overview of what's happening.

NYT is writing to audience that just wants to know whether they need a fancy new router or whether they should stick with their current one for a while longer. That's it.


I'm not debating the audience, I'm debating the qualifications of the person writing.

I know Wendell is the real deal because Dr. Ian Cutress wouldn't have had him on AnandTech livestreams if he wasn't. I also know Wendell's the real deal by watching Level1Techs YouTube videos. I also know he's the real deal because he literally builds out networks for companies for a living (Wendell Wilson Consulting in Kentucky). I don't know shit about this NYT writer. His credentials aren't listed on the NYT website. Does he have a degree in business information systems? Electrical engineering? Computer science? I don't know. That's the whole point.

I don't want to hear heart health advice from a "health journalist", I want to hear what an actual cardiologist has to say. Same with a "tech journalist". No thanks, I'll trust the opinion of someone who builds networks for a living when it comes to my network equipment.

In the past, finding people with these qualifications was difficult, and then you had to find time to liaise with them. Now you can find some of the best people in the world and gain their insight just from reading or watching YouTube.


Back in the old days there were media outlets for the general public and the specialists. One just has to be pointed to the right direction. I don't think today's media landscape is that different - the professional youtubers are just today's trade journals.

Though there is definitely a much larger barrier of entry when it comes to video because product is much more involved than just writing an article. I've noticed over the past few years was that pros from the few non-tech hobbies have been trying to get into video and many are clearly struggling. I often wonder how much talent is out there that we don't get to see.


You seem convinced Wendell is well known and authoritative, but I've never heard of him, any more so than the author of the Times piece.

You should consider that not everyone shares your experiences.


> I'd rather wait for Wendell Wilson to do a video and I'll watch it on Level1Techs channel on YouTube.

I watch a ton of tech Youtubers and I have absolutely no idea who those folks are. The world is a lot bigger than you imagine.


I know exactly how big the world is, thanks. The guy personally built a new workstation for Linus Torvalds and sent it to him.

Maybe the world's a bigger than you imagine.


I'm not sure you get my point.

Your internet celebrities are not as famous as you imagine and thus not references the layperson would know of let alone trust.

A major newspaper, despite its defects, is much more of a known quantity.


Wendell did a video with Linus Tech Tips a few years back, and they’ve collaborated a few times before and since, too:

https://www.youtube.com/watch?v=SsgI1mkx6iw


network chuck too, was my first video of him I watched, and probably most interesting too, but that's probably personal thing.

https://www.youtube.com/watch?v=8cmmVEoftEM


I can’t find any appearence or mention of Wendell in that video, nor could I find any other video by the same person featuring Wendell.


Most people aren’t technical experts, so this has a much broader appeal using language and scenarios that’s more familiar to everyday people.


That's not my point.

My point is why should anyone listen to this guy. None of his qualifications are on display. Does he build out WiFi networks for a living? "He works for the New York Times" is not a qualification anyone should give a shit about after how they've conducted "journalism" since 2015.

I want to hear from experts, not journalists. Maybe articles wherein journalists consult with a range of experts and then convey those experts' opinions to me, but I definitely don't care what a guy with a degree in journalism who happens to love tech has to say. I want to hear from people from work with these technologies from day to day.

If you want to know about heart health, you ask a cardiologist, not a health journalist.


I think the issue your ideal world of communication runs into, is that experts can't dedicate all of their time to public outreach. That's where journalists come in. If you want to know why your chest hurts a bit on the regular, you talk to your doctor. If you want to learn a few interesting facts about a heart condition, you read an article written by a journalist.


For a network that's already working fine, it's only going to make marginal improvements.

Who's going to spend $230 on a new router for slightly faster lightbulbs?


This comment made me laugh. My light bulbs turn on at the speed of electricity right now.


thx for pointing that out, you two made my day!


The fact that lightbulbs took seconds to turn on suggests something was very wrong and he's getting much more than marginal improvements.


Most likely due to slow ARP resolution. My custom firmware power switch would take a long time to respond until I turned on gratuitous ARP broadcast.


Anecdata, but I recently went to wifi6 and even though I didn't notice until now, I'm pretty sure our wifi lights respond quicker than they used to. It's maybe only a second or 2, but I remember being annoyed at how long they took sometimes before, and I haven't felt that way since I got this new router.

However, the old router was a Google mesh and 1 of them definitely died slowly, so that could have been causing the delay for a while before it fully died on us.


If your lights take seconds to turn on, invest in better technology. Lutron Caseta switches take milliseconds to turn lights on.


Sounds like a placebo or he had terrible wifi before. Alexa and the old bulbs wouldn't be running Wifi6 so its just a newer router running in backwards compatible modes.

The only Wifi6 device the author owns is probably an iPhone.


It does. I have a hard time believing Wi-Fi itself could cause such a large swing in latency, but perhaps all those back and forth round trips do add up? Is WiFi 6 known for major latency improvements? Or is this just related to less congestion?

It would be interesting to compare it to Ethernet usage of these devices.


Yes, WiFi 6 has a lot of major improvements to reduce latency.

Intel has a good write up here:

https://www.intel.com/content/www/us/en/gaming/resources/wif...


> seconds to turn on a light bulb

What a time to be alive! We truly are living in the future!


We used to have to take several seconds to walk over and back to a light bulb to turn it on. Now it's just as fast but we don't have to even stand up.


I'd be curious to know whether that's a wifi problem or a "your IoT devices are run by very small computers running very unoptimized software" problem.


I upgraded to a Wifi 6* router last month. We have Gb service. I could never get more than a few hundred Mbps from any of my devices on the 5 GHz bands but with ax I can get over 800 Mbps. I only have two ax devices and their performance is startlingly different (one gets about 400 Mbps but the other gets over 800).

I do wonder how much of the issue is channel contention from my neighbors and how much is actual improvement.

* Not mentioning brands because I don't want to look like I'm shilling. Anyway I only had a couple of dozen clients to test with, and only two (before and after) routers.

Edit: I had "K" and "G" where I should have had "Mbs" -- thanks jeffbee and gratin for pointing these typos out.


> but with ax I can get over 800 Gbps

Can I assume you meant 800 Mbps?


Thanks, an absurd typo; fixed.


You were getting a few hundred kbps per device on what kind of network? That's not even enough for CD-quality digital audio. That wouldn't have been acceptable performance even in the 1990s before 802.11.


Thanks, an absurd typo; fixed.


Most likely you upgraded to an ap with working dfs detection and it just hopped on a better channel...


That is amazing.


Two typos made it extraordinarily amazing; if you were politely referring to those: I edited the post.

However, even with the correct units I am astonished and gratified with the improvement.



> I usually have more than two dozen internet-connected devices running, including smart speakers, a thermostat and a bathroom scale. That appeared to make my home an ideal test environment for Wi-Fi 6.

mu-mimo (802.11ax) requires client side support, since it's a way to make clients use different channels than the main channel they would have used pre-802.11ax. i'm skeptical this test actually tested mu-mimo, that the wifi 6 router was doing anything new or better.

not impossible though. it helps that some pre-wifi 6 devices (802.11ac, 2014) started getting mu-mimo/802.11ax support but only on 5GHz bands. so some not fresh devices may have support. i believe anything with 802.11ax support on both bands qualifies as wifi 6. haven't found good links to support this though! i wonder when the first client 802.11ax devices emerged!!

the Wikipedia reference is pretty solid: https://en.m.wikipedia.org/wiki/IEEE_802.11ax


Is anybody excited about Wi-Fi 6?

I'm just not seeing a compelling use case for many residential users to upgrade given the number of legacy devices that don't support this protocol. More importantly, the impending release of WiFi6e routers that open up 6 Ghz makes me want to wait even more... That is a huge potential bandwidth increase.


I am not excited, but I am happy that the 2.4ghz band finally got an update. 802.11n needed to catch up to ac.


Yep, not excited for Wifi6, i'll wait for wifi6e to upgrade. My AP-AC-HD works great.


5 GHz to 6 GHz is 6/5s faster bandwidth...


I bought a house and recently bought a router for our Comcast service. I wanted to “future proof” so I bought the ASUS WiFi 6 Router (RT-AX3000).

Before at our apartment a little further down the road with the same Comcast service, we were always flipping WiFi on and off on our phones and tablets to reconnect because of hanging connections. I was blaming Comcast, but it’s now clear to me that it was the router and not the connection itself. Our new router has never dropped my connection and I get a signal throughout the house. Streams start instantly and downloads are as fast as at the office. I can work from home via VPN and it’s indistinguishable from sitting at my desk on a wired connection.


I’ve an unusual layout in my house and WiFi placed in the middle of the home would not reach either side of the long narrow floor plan. I tried various mesh options including ubiquitis prosumer amplifi but that proved very unreliable. Before going full enterprise grade I tried the Linksys Velop WiFi 6 that has 2 Access Points. It has been tremendous thus far. I have WiFi well out into my yard(live in country area on acreage) and aside from occasional hangs the system has been stable and covers the entire home.


For me, 2.4 ghz has worked very, very well for penetrating pesky walls full of HVAC/washer/dryer/giant metal beams. I keep my office devices either plugged into ethernet or connected to 5 ghz where they're in clear view of my router. Especially if you're not right on top of other apartments or anything and don't require >500mbps download speeds, any decent router should serve you well on the 2.4 band.


I'm skeptical of his results, or at least that his results had anything to do with Wifi 6:

For one, my Amazon smart speakers are now more responsive. In my bedroom, I ask Alexa to control a pair of internet-connected light bulbs. With the older router, whenever I said, “Alexa, turn on the lights,” there was a delay of about two seconds before the lights turned on. Now it’s less than half a second.


In other words author is a Hack Fraud. Reminds me of The Verge PC Build Video tutorial done by another "tech journalist" while wearing a _wireless anti static bracelet_ https://youtu.be/0vmQOO4WLI4?t=46 https://www.youtube.com/watch?v=M-2Scfj4FZk


Much of the improvements you get with upgrades are often just redoing the configuration correctly. A two second pause sounds like a timeout of some sort.


Yeah. The garage door opener the author mentioned supports only 2.4Ghz networks


Surprised that there's no mention of Wi-Fi 6E, which is rumored to be in the Apple phones coming in the fall. Having the 6GHz band as well as 2.4GHz and 5GHz seems like you'd get a year or two of having lots of room for your devices.

With work-from-home and online school, I had to move an Orbi satellite so Wi-Fi 5 was more evenly distributed. 6GHz will have even less range than 5GHz but having it and the other changes in Wi-Fi 6E could make a huge difference for offices.


> Wi-Fi 6 reduces congestion by directing traffic. There are now multiple lanes: car pool lanes for the newer, faster devices and a slow lane for the older, slower ones. All of the vehicles are also full of people, which represent big batches of data being transported over the network simultaneously.

Nice car analogy; how will that play out in a crowded condo building, where neighbors around you have multiple routers with umpteen "lanes"?


I’ve been looking for a Wi-Fi 6 router with 10GBe ethernet ports too.

Its kind of silly to have wireless that is faster than the wired options. One use case is that I want to store larger files on a NAS but I don't want throughput to randomly drop such as during large image backups and restorations to a cloud server.


WiFi specs refer to the maximum theoretical link-layer rate in perfect conditions. The actual amount of data (not link-layer bits) transferred is less than the headline number.

In the real world, 1G ethernet isn't going to hold back a WiFi 6 router. The only exception might be a multi-radio WiFi 6 AP that is within several feet of WiFi 6 clients on both radios.


was going to say this.

feels kinda weird that wifi marketing speek is taken at face value two decades into the bullshitting with theoretical maximum agregates.


I'd be surprised if it is much of a restriction in real world use cases. Accessing my NAS over 866Mbps WiFi[1] caps out around 60MB/s, the same devices over gigabit ethernet hit around 120MB/s (the NAS delivers around 650MB/s over a 10Gb link).

[1] tested with multiple client/router devices, in close to ideal circumstances (few feet line of sight between devices, no other devices active on the network)


802.11ac from Google Wifi maxes out at around 300 mbps for me (on gigabit internet. for one device). Not sure on your setup, but it does matter how many streams your Wifi network adaptor has.

The advertised speed of your router usually doesn't mean much, as it usually implies the maximum concurrent speed available, rather than how much one device can use.



The AXE11000 and RAXE500 both have 2.5G ports, at least.


And the EAP660 HD.


I'd be happy to just have wifi that didn't lose its mind when you run the microwave.


Don't you just need any wifi at 5 GHz for that? 802.11n (aka Wi-Fi 4) has been around for quite a while.


Your microwave is probably leaking. I've never had issues with mine and it's a cheap one, so if even a cheap one can insulate against leakage then yours is most likely faulty.


I have two microwaves, completely different designs, and either one running will knock out the wifi.


Try running both at the same time, maybe deconstructive interference will not harm your Wi-Fi


I'm actually still surprised at how expensive Wifi5 routers are. I still see "top of the line" wireless routers which still use 8 year old technology.


Pantiently awaiting https://www.netgear.com/home/wifi/routers/raxe500/ to actually be purhacasable so the 40-60 other wireless networks near me just "disappear". Already have all my non-wired clients upgraded with 6E m.2 chips, phones aside




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: