[greatgib]

<<A special discount code has been emailed to OnHub users only, for 40% off Nest Wifi on the Google Store.>>

Looking at that, I'm wondering who is stupid enough to buy a new device from them like this. It is like 'bite me once, bite me again please.'.

'eh, we just self destroyed your device, because, fuck you, but here is a coupon to buy a new one from us that we will also kill in a few years' ...

[yjftsjthsd-h]

I dunno, I'd have to do a TCO analysis; if they give you 40% off, and you just assume it'll die after X years but be fully supported for that time and then you upgrade to a faster unit with next-gen wifi or whatever, it could still work out to be a good deal. Like... not my cup of tea (TCO is even nicer when the unit's lifespan is indefinite), but it's a plausible tradeoff.

[gruez]

>and then you upgrade to a faster unit with next-gen wifi or whatever

Do people actually need next-gen wifi? My impression is that 99.9+% of people would be well served by a mid-range 802.11ac router (eg. AC1750[1]) from years ago, because the most bandwidth intensive thing they do is watch 4k netflix (~25 Mb/s).

[1] https://en.wikipedia.org/wiki/IEEE_802.11ac-2013

[admax88qqq]

I'm sure people said the same thing about 802.11ac when everyone was on n, or n when everyone was on g.

Faster is always better. New 802.11ax also improves contention between devices.

[ineedasername]

I have ~7 year old bog standard wifi router that has more than enough speed to saturate my 200Mbs downlink. Up to 7 or 8 devices online at a time when all family members are home, and there's never a contention issue. I can hog the connection w/ a huge Steam download, and then when they kids start streaming netflix to the TV it's still all shared nicely with me getting a reduced speed as the stream kicks in, everything still working perfectly & taking full advantage of the 200Mbs speed.

That's not meant to refute you: Sure, faster etc. is good. But I'd be pretty angry if I was told it was being bricked by the manufacturer under the conditions that Google is citing when everything still works perfectly well. There's no reason at all that Google has to disable management within the app.

And since I have a new finished basement that doesn't get the WiFi signal very strongly I am considering an upgrade to a modern mesh setup, and was heavily considering Nest Wifi. Now? Nope nope nope nope nope nope nope.

[vel0city]

I'd highly recommend looking at just getting a wired AP to put in the basement and have it ceiling mounted. It really depends on how your house was designed, but it could be relatively easy to drop a cable down from an above floor to the ceiling. You'll get much better performance having a wire carrying the signal than trying to do WiFi mesh setups, especially when you're trying to cover an area that already has terrible WiFi performance.

[mook]

In my experience even Ethernet over powerline is better than using WiFi mesh, since there's still a wire, even if it's noisy. That is, it provided a more stable connection, even if the peak speed is not theoretically as fast.

Your mileage may vary.

[thetinguy]

You live in a house. In an apartment there are 50 other networks within a 100 feet of you.

[ineedasername]

You actually don't know my residential environment. I don't live in an apartment with 50 networks, but I can see ~30 SSID's right now. On a semi-regular basis I benefit from going into my router config to choose the channel with the least congestion. For me, that's not worth spending $$ to upgrade when speed & latency are not otherwise an issue. I don't begrudge people who decide to upgrade to get that incremental boost, but I don't think it should be forced on them. Because I'm not claiming there are no benefits to upgrading. I'm saying that people who have lived with their current WiFi setup-- hiccups & all-- for years shouldn't be required to pay more if they're satisfied with their own status quo.

[gruez]

>I'm sure people said the same thing about 802.11ac when everyone was on n, or n when everyone was on g.

You'd probably be well served by 802.11n right now. Take a typical setup of 802.11n with 2x2 MIMO: theoretical speeds would be 300Mb/s. Halve that for "real world" performance, and you have speeds of 150Mb/s. That's more than enough for multiple streams of 4k netflix. Looking around it looks like most 802.11n adapters of the day (I checked 2011, the spec was released in 2009) only support 2.4ghz, so spectrum congestion might be an issue in high density areas, if we halve the speed again to account for this, we get 75Mb/s, which might be slower than your internet access, but is still otherwise usable.

Based on this, I'm thinking the prudent strategy is to upgrade every other wifi generation.

[lmns]

802.11n with 5 GHz is okayish, but using 2.4 GHz can be brutal. It's not just the bandwidth, the whole frequency band is just bad. My latency regularly goes through the roof and voice calls are noticeably worse than with 5 GHz wifi.

802.11ac mandates 5 GHz, so it's a reasonable baseline. I don't even care if it's SISO, MIMO or whatever as long as I don't have to fall back to 2.4 GHz.

[wtallis]

At the moment, I would definitely choose an older 802.11ac device with mature open-source drivers over any 802.11ax device. Especially since last I checked, the only affordable 802.11ax routers were limited to 2T2R, so they would be a step backwards in peak throughout for some client devices.

[diamondo25]

Latency is a thing, and newer networks have lower latency or are affected by other clients being slower (eg MIMO and improved channeling), though other ethernet devices can still have a big effect (eg remote server latency, dns latency, retransmissions).

[wtallis]

Latency is exactly why I'd prefer a last-gen radio with open-source drivers. An 802.11ac router with the latest open-source drivers (ath10k and mt76) will give you better real-world latency, because there are open-source developers working to optimize for the metrics that actually improve real-world user experiences rather than seeking the largest achievable throughput numbers on naive speed tests.

802.11ax on its own does nothing to address the problem of wasting too much airtime serving a client at the extreme edge of your AP's range where neither 802.11ax or ac transmission rates will work. That kind of problem must be tackled with better QoS strategies that are not inherently tied to any particular generation of the WiFi standards, but in practice can only be developed and tested for chipsets that are sufficiently hackable

[baskethead]

Faster is not always better.

I upgraded to 10Gbps for my network and I barely use above 1Gbps, what I originally had.

I upgraded my Wi-Fi to wifi6 and I can’t tell any difference at all, despite having around 40 wireless devices on my network. All our phones support it and I literally don’t notice anything.

We may be getting to the point where more speed is not as important.

[nitrogen]

We may be getting to the point where more speed is not as important.

To an extent this is true, but part of this is just how long US broadband speeds stagnated, especially upload speeds, killing off any product ideas that required high upload bandwidth. I'm sure that over time we'll find uses for more speed.

[vel0city]

> I upgraded my Wi-Fi to wifi6 and I can’t tell any difference at all

How many WiFi 6 clients do you have though? If all (or even the large majority of) your clients are 802.11ac clients, you're not going to notice any real differences.

The biggest differences are going to be spectral efficiency especially when it comes to many devices chirping small packets. So think having lots of IoT things around the house while you're doing VoIP and gaming stuff. Things like streaming video won't really see an incredible difference most of the time. Streaming video usually has at least several second buffers and isn't actually constantly clogging the band, its brief blasts of several seconds worth of data at a time.

[pantalaimon]

> I'm sure people said the same thing about 802.11ac when everyone was on n, or n when everyone was on g.

Certainly not - 802.11ac felt like the biggest leap, 802.11n was decent for browsing, but when starting a download I would try to switch to Ethernet for it not take too long. With 802.11ac I didn't have to bother anymore as usually the difference was negligible.

[jjav]

> Faster is always better.

For wifi faster isn't always better if there's nowhere to go. That is, if the uplink from home isn't itself faster.

[jordanthoms]

I ran a Asus ac-68u at home at home for years and was generally happy, but the speed bump from upgrading to a ax-86u has actually been very substantial even for ac devices.

BTW I’m using a mikrotik router, but I’m not sold on Ubiquiti for home use even though I’ve deployed plenty of them at our office - my impression is that while the management on them is nice and they may be better for handling large numbers of clients, the more high end home aps have long range external antennas and seem better at providing high speeds to a small number of devices. It’s hard to find many objective tests of that though…

[YokoZar]

> Do people actually need next-gen wifi?

Streaming VR to a wireless headset (SteamVR on Oculus Quest) is generally considered to require Wifi 6, though that is admittedly a niche use currently.

[yjftsjthsd-h]

The question isn't whether people need it, question is whether people were going to use it as a reason for upgrading anyways

[basch]

is Bandwidth the only factor?

What about 30-50 smart home devices? Every bulb or switch, water sensors, etc. Some of modern equipment is load balancing old legacy stuff better, no?

[gruez]

I don't have an IOT house, so I don't know how well they'll cope. That said, most IOT chips (eg. the famed ESP8266 and most variants of its successor, ESP32) seem to only support 802.11n, a standard from 2009. Therefore any benefits are probably not going to be gained by upgrading to an 802.11ac router. Newer routers might have beefier cpu/ram, but IOT devices are pretty low bandwidth, so I doubt they'll be taking up much cpu/ram to handle them.

[Tijdreiziger]

For smarthome, I would recommend Zigbee devices. That way, you limit your attack surface to just your Zigbee gateway, rather than every single light bulb in your house, with the added benefit that those don't clog up your Wi-Fi.

We have the Philips Hue gateway at home (which has limited support for other brands, but IKEA Trådfri bulbs work), but you can also get a Zigbee USB stick/Raspberry Pi HAT, install Home Assistant or deCONZ, and go wild without being tied to any vendor.

[prmoustache]

Why would you want to cram 50 so called "smart" devices is the real question. Is your quality of life improved by having your lightbulbs and switch connected to the internet. Why do you need sensors at all?

My home is completely analog, like I even have to operate my windows blinds manually. I have lived in a totally connected and electrified house and I didn't view any improvement in my quality of life. Only downsides when the power grid had an issue.

[5e92cb50239222b]

Well, if you want an example, I have about about a dozen PM and CO₂ sensors spread around the apartment, and some outside of it. I live in an extremely polluted area and it allows me to make decisions on when it's (relatively) safe to open windows (or even inside doors if it's really bad outside).

PM 2.5 may be in the range of a couple micrograms/m³ in one room (with air purification) and 100× that in the other adjacent room, so it makes sense to have sensors everywhere.

Still very far from 50, though.

[pydry]

My quality of life is improved by heating that can be turned on remotely (while im on my way home) or on a timer (before i wake up).

Lightbulbs not so much.

[vel0city]

You don't need to have it internet connected to have your thermostat on a timer. My home has responded to heat or cool the house based on when I wake up, when I leave, when I come home, and when I go to bed for well over a decade all without needing an internet connection to do so. My thermostat doesn't go dumb just because the internet went down or because it became unsupported and got bricked.

The only thing I wish it would do would have better multi-zone temperature and humidity sensing to know to turn on the circulation fan when the edges of the house get too hot/cold compared to where the thermostat is. Even then that doesn't require the internet, it could be done with cheap 433MHz temp/humidity probes running on button cell batteries for years.

[greatgib]

You TCO is negative in all cases. Because you had something that was working enough, otherwise you would have changed it. But now you need to spend money to change. Even if for a good price.

Also you have to take into account that you pay but not have 2 working devices in the end (old and new). Just one.

Sure, if the new one was providing significant value, there could have been an interest. Like if you network was saturated. But that would be surprising because currently 8/10 years old ac wifi routers are more than enough for most households usage.

Also, what is interesting is the Stockholm syndrom: some comments here are saying that it was probably old enough to be changed, so not so bad. But in that case, did you think about that, that you would probably have to change your router for a new one before receiving the obsolescence notice from Google?

[vel0city]

> currently 8/10 years old ac wifi routers

802.11ac was released in 2014, only 7 years ago. You wouldn't have a 10 year old AC router, and really probably wouldn't have an 8 year old AC router quite yet.

[ineedasername]

I would only agree with that if Google fully published their support roadmap ahead of time so that I could know what I'm buying. Especially in light of the $ at any given time: $300 when there's 6 years of support ahead is fine, but it's not worth $300 anymore when there's only 4,3,2 years of support left. How can you possibly do a TCO analysis without that information?

They at least do this for their Pixel phones.

[zmj]

I probably will. The OnHub has performed substantially better than the other consumer wireless routers I've used, and after 5 years it's due to be replaced anyway. Might as well take the discount.

[ineedasername]

Why is it due to be replaced? Has your bandwidth speed substantially improved to the point that OnHub can no longer keep up? Because otherwise it sounds like you're happy with where you're at. There's no reason to replace a well-functioning device that does everything you need just because it's reached the end of the ever-shortening consumer electronics lifecycle. My own router is about ~7 years old, I have twice the devices connected as I did when I purchased it, and it still handle my 200Mbs broadband just fine among all devices in use.

[acdha]

WiFi isn’t that simple: every generation makes substantial improvements in things like how they handle contention with multiple clients or interference. Since most of us have more than one device active these days, that is noticeable — especially if you have devices which also talk to each other rather than exclusively over the internet.

When I replaced my 2016 OnHub + Google WiFi mesh with an Eero WiFi 6 system, I saw across the board improvements in latency and bandwidth from almost every client (faster ones more than doubled). We also lost the sporadic hangs for iOS devices which Google never fixed for totally innocent reasons.

[ineedasername]

I think that what is simple for me is that I don't make those judgements based purely on age. If I'm not having noticeable problems with a 6-7 year old router then I'm not going to write off forced obsolescence as "well it was time to upgrade anyway". I already get the full 200Mbs speed that I pay for and 15-20ms ping & low jitter. Upgrading isn't going to do a lot for me.

Everyone is able to make their own determination on that sort of thing, when they should upgrade. Or at least they used to be able to do that for this product class. Google is now baking in expiration dates that aren't even published at the time of purchase.

[acdha]

I’m not defending Google but almost everyone I know who replaced a 6+ year old router then spends a month saying they should have done that sooner because the differences were noticeable. Given how heavily most people use WiFi now, spending something like $0.50 per day for your entire family seems like far from the most pressing area to economize.

The other thing to consider is how the rest of the market compares: a lot of people have routers which still work but are no longer secure, so this entire field seems right for a legal requirement of, say, a decade support period and/or mandatory recycling.

[ineedasername]

far from the most pressing area to economize

$0.50/day seems like a small thing until you're strapped for cash but all of a sudden have to pay $500 all at once to replace your Google mesh network hardware.

That aside, as much as this particular example frustrates me, it's not my primary concern: It's the overall trend of forced obsolescence taking yet another step forward and increasing issue of e-waste.

Mandatory recycling is also problematic. It could mean that products that might have ample community support (e.g., via OpenWRT) would still be illegal, and in general would take away a user's right to support & maintain their purchases. It is also something that would be ripe for regulatory capture.

Separate from those issues is the consumer's ability to make an informed choice: A product with a potential expiration date should be required to market it as such. Google does this with Pixel phones; hopefully after this they will begin doing it with their other products as well, and I think that in general it should be required: MS does this with Windows, plenty of other vendors do it, there's no reason it can't be a universal requirement as part of consumer protection laws.

[causi]

Did you notice those improvements or did you just do benchmarks to see if they improved? Not everyone's needs are the same of course, but I have no complaints about my seven year old Nighthawk R7000. Latency-sensitive devices like game consoles are ethernet, phones and laptops are 5GHz, and everything else is 2.4GHz.

[prmoustache]

I think this is a case of first world country arrogance of putting perfectly working things in the landfill just for the heck of it.

[acdha]

I found it noticeable for anything latency sensitive - conferencing, Terraform/SSH, etc. (losing Google’s iOS hangs was extremely noticeable, of course, but that’s for a different reason)

The key thing to remember is that there’s considerable variability across different peoples’ experience. If you live in a widely spaced suburban house, you probably don’t need to worry about interference the way an apartment dweller does or even someone in a city where the neighbor’s property starts a lot closer. The layout and materials used in your house similarly have a big impact on whether your devices are operating on 5GHz close to a base station or 2.4GHz further away.

Since you have your latency sensitive devices on Ethernet, you probably are fine with an old AP. Cables are definitely better but not everyone has permission or a suitable way to run them (my house built in the 1930s did not include spare conduit).

[ajdude]

I’m guessing that Google has datamined everything that it could from all of the users who would get something like this, so there is no point financially in supporting it.

[cube00]

Make sure you check out some of the other "pro-sumer" lines first. I moved from a flaky consumer router which dropped Wifi all the time to a Ubiquiti which has been rock solid.

I'm not recommending Ubiquiti specifically given recent events but just using it as an example of a prosumer brand, even a few years old now and it still gets firmware updates, where as my old consumer router would be lucky to get any updates at all.

[robocat]

> Make sure you check out some of the other "pro-sumer" lines first

Unless you enjoy networking as a hobby, I would worry with that strategy that there is a sizeable risk of wasting your time with unreliable devices, with ongoing maintenance work, or having to rework your network in a few years.

My current solution is to buy Unifi UAP-AC-LR (~120USD) and configure it as an access point using the “Unifi Network” app from my phone, and hopefully never touch it again. I have done this at friends, and fixed their WiFi woes, without requiring much of my time (occasional complaints that the WiFi isn’t working, but not due to the device*, just ISP or router issues). Easy to plug the AP into a new router if you change ISP or move houses.

* Well, one device just stopped working with a hardware fault: I think due to being installed in a very hot area. I haven’t had software issues or flakiness. Flakiness is my previous experience of prosumer devices and what I most want to avoid.

[cavisne]

Doing exactly this works well, but the other bits of UniFi kindof suck. The desktop software is flaky and I’ve seen some really unreliable firmware updates. The non AP hardware is really overpriced as well

[slenk]

That works great if you can run PoE everywhere.

Do they have a unifi version that I can plug in to a wall outlet without having to upgrade everything to PoE too?

Unless you have PoE you can't add a unifi AP for $120 USD. And getting into PoE isn't cheap.

[seanp2k2]

Some UniFi APs used to come with PoE injectors. I’m not sure if they still do, but it looks like they still sell them for $8:

https://store.ui.com/products/u-poe-af

If someone is willing to spend a bit more, the 8 port managed GbE switch with 4 PoE ports is $110: https://store.ui.com/products/unifi-switch-8-60w

That said, they’re coming out with a “UniFi Dream Router” for $80 (might be more once it goes GA, it’s in beta now via their Early Access program) that has the wifi, gateway, and management all in one device https://dongknows.com/ubiquitis-wi-fi-6-unifi-dream-router-u...

If someone is interested in going all-in on UniFi, their best solution is coming in the form of the Dream Machine Pro SE that can support their camera platform (Protect) along with Network and includes a 10Gb SFP+, 2.5GbE, and a PoE 8-port switch which should be fine for most houses (e.g. 5 wired PoE cameras + 3 PoE APs).

I’ve been into Ubiquiti ever since the OG EdgeRouter, so my current setup is a bit more complex. I’m not happy that they’ve ended support for UniFi Video that I ran for years on a NUC with Ubuntu that also hosts other small home automation stuff and Jenkins. I finally caved and got a Cloud Key 2+ to run Protect, which only runs on their hardware, even though it’s just 64-but ARM. At least it’s fast, I guess. It would be great if they had any real competition in this space, but all the other DIY and enterprise options really suck for cameras. MikroTik is fine for networking gear, and with the prices I’ve been seeing on some wifi mesh systems and even “gaming” routers, people could be getting Aruba Instant On or Ruckus Unleashed systems that would be infinitely better than the junk that they’re passing off as “gaming” routers these days.

[altantiprocrast]

When you buy a single Unifi AP they include POE injectors (i.e. a power brick that injects power into ethernet).

[tinus_hn]

You don’t need to ‘run’ PoE everywhere, it just provides power over the cables you are already using.

[mrkstu]

Yep, Unifi + bridge mode for the win. Make wireless as dead simple as possible and move your intelligence to your router/firewall.

[ineedasername]

>Re: Ubiquiti

Investigators say they were able to tie the downloads to Sharp and his work-issued laptop because his Internet connection briefly failed on several occasions while he was downloading the Ubiquiti data. Those outages were enough to prevent Sharp’s Surfshark VPN connection from functioning properly [0]

Yeah, surfashark takes an extra few minutes to load after you start your machine. Maybe he got over eager. I wonder if he has a good lawsuit case for this "Your honor, I was trying to do crime but the failure of the defendant's product resulted in a life-ruining indictment"

[0] https://krebsonsecurity.com/2021/12/ubiquiti-developer-charg...

[sosodev]

Yes! I’ve been using a Microtik router and Ubiquiti access point for a couple years now with zero issues.

[joshu]

I replaced Ubiquiti with refurbed Ruckus stuff and am extremely happy.

[tw04]

I would suggest taking a look at synology. The 2600ac is going on 6 years old and about to get the latest feature release (1.3) with no immediate plans to eol it. And unlike Google they have a long history of LTS.

[rpdillon]

This is precisely what I did. My OnHub was having issues about 18 months ago, so I did some research. I was mostly interested in getting a browser-based interface back, push security updates from the vendor, and buying from a company that had a good track record of treating customers well and supporting their products. I picked up the Synology and have been very impressed. It got my family through the pandemic scenario of four simultaneous video conference streams with 30+ participants each with flying colors. This is my third Synology product, and I have an extremely high opinion of them. The other two products are NAS products, and both have been just as bulletproof for me.

[pde3]

How do you find the RF performance? For me the OnHub was vastly better than other home wifi gear I've used, but would consider more trustworthy replacements if they have comparable performance...

[tw04]

I guess I don’t have a ton to compare it to. Installed for the neighbors and they get signal from the office in the front of the house to the pool out back, so good? Before that they had a cambium router running DD-wrt that didn’t have anywhere near the range.

You can also buy multiple, they do mesh networking either wireless or wired. They also just announced a 6600 that supports WiFi-6e.

[worik]

Is five years enough?

For you maybe.

I think it is nothing like enough. My guitar amplifier is about twenty years old.

My house is thirty and will probably make it to 300 (climate catastrophes wiling)

I do not buy Google products, if I did I would stop.

[jkelleyrtp]

Well, electric guitars have been around for over 80 years and houses even longer.

This whole wireless internet thing is pretty recent and is getting increasingly better every year.

[Arcanum-XIII]

I have some power tool from the 60, with a deviation of precision about a 1/100mm from manufacturing time, while abused in factory. Heck, I just got a new drill from 1959, with the same. They were also “recent” design at the time…

[prmoustache]

But if you are living with the current performance just fine, why do you have to seek a faster one.

As long as I can work, play 2 concurrent movies while having videocalls at the same time, I don't see the point upgrading just for the sake of upgrading. Unless the security of my devices is at risk.

[pantalaimon]

> after 5 years it's due to be replaced anyway

Are people really replacing their home network infrastructure every 5 years?

[harha]

The best move is to add cables where possible. I’ve got a slightly upper scale consumer mesh network (with Ethernet backhaul) for anything that moves, everything else is connected via cables (have to admit that I only had to draw one cable from the fiber outlet to the other side of the room where an Ethernet outlet was ready) and gigabit switches, both of them at a very low cost. I get almost the same speed and latency at any wired device as I do at the router, even though the cables going to the rooms are 10+ years old.

Unless you’re living out in the wild with no devices to interfere, speed and reliability will be well worth the effort and save a lot on ineffective Wi-Fi gear.

[dsr_]

Very much agreed.

If a network-attached device is always in one place and it has an ethernet port, it has a cable.

If it's always in one place and it doesn't have an ethernet port, there's a cable to the wireless point in that room or the next.

5-port gigabit switches are about $20 each. A 12-port gigabit switch anchors the whole thing. None of them take configuration.

This is harder but not impossible for people who live in apartments; white cable run along the edge of the ceiling or along the foot of the wall is a good bet.

[ocdtrekkie]

Definitely this. I use any time where I have to open a wall as an excuse to put Ethernet in it. Bathroom getting renovated? Great, use the opportunity to run Ethernet up to the second floor. Garage has a drywall puncture with moldy insulation behind it? Great, run Ethernet.

The upside here is adding Ethernet has been very low cost for me: Literally the cost of the cable and the keystones. Downside is I've been here like two years and my Ethernet runs are still somewhat random/piecemeal.

In one case, which has performed surprisingly well for basically the whole two years: I used an existing coax run with a pair of Motorola MoCA adapters, which provides a gigabit connection from my basement to a room that's particularly hard to retrofit a connection to.

Wi-Fi is for guests and smartphones. Ethernet is for life.

[jjav]

> The best move is to add cables where possible.

Can't recommend this enough. WiFi is great for mobility but it's just not that reliable. For anything that doesn't have to be moving, pull some ethernet to it and be happy for the long haul.

[bonoboTP]

I use the same Linksys WRT54GL since around 2006 and it never drops the connection. I flashed dd-wrt on it back more than a decade ago and it "just works".

[gjsman-1000]

Heck no - ask an ISP, they keep them boxes in homes till they drop.

[joshuamorton]

5-6 years seems to be about the time between wifi revisions (1999, 2003 (g), 2009 (n), 2013 (ac), 2021 (ax/6))

[jakogut]

> Looking at that, I'm wondering who is stupid enough to buy a new device from them like this

It seems this is more common than it should be. Apple users who complain of new products lacking features older products had, but who continue to buy those products come to mind.