TR Forums

Anyone Set up a home network using Ubiquiti APs? I have a question if you have, What switch are you using for the PoE? Did you stick with Ubiquiti ones or use something else? Also do you use a cloud key?

My goal is trying to save money and if I can go with another switch that provides PoE to the Ubiquiti APs that would be great.

I've got a Unifi AP AC LR and I'm just using the PoE injector that came with it. I have a Linux server box so I just loaded the controller software on that.

If you are going to run it off a PoE switch then watch out for which standard the switch supports. Different models of Unifi APs use different standards of PoE, see the table at the bottom of this page.

I have 3 Unifi APs, but I used the supplied injector for each one with my pre-existing regular non-PoE switches.

As notfred said, you *ABSOLUTELY* need to watch out for the different standards for PoE. Check his link out.

So, given that you get an injector with each AP, you typically are not "saving" money by buying a more expensive PoE switch unless you have a complicated wire/power situation or are paying someone to do like a professional permanent installation etc...

I don't use a cloud key either. I just use the regular java-based controller software (which you don't even need to leave up and running in order to use the APs after they are configured/updated)

This would seem to be the cheapest option (I am very cheap lol) in the generic scenario.

So I am going to use the Lite(s) (802.3af/A PoE & 24V PoE) I haven't purchased them yet so you guys are telling me each one comes with its own injector?

So I can go from router/switch -> injector -> AP?

Hmmmm....but that going to be very messy if I use an injector for each AP. But if I'm going on the cheap then that is probably the best. FYI, I have had the house wired for 5 AP drops.

I have 2 AP-LR's and an USG, with the software running in Docker on my Synology NAS, working quite well.

I just used the included injectors, as I already had a 24 port switch.

tanker27 wrote:

So I am going to use the Lite(s) (802.3af/A PoE & 24V PoE) I haven't purchased them yet so you guys are telling me each one comes with its own injector?

Yes.

tanker27 wrote:

So I can go from router/switch -> injector -> AP?

Yup.

tanker27 wrote:

Hmmmm....but that going to be very messy if I use an injector for each AP. But if I'm going on the cheap then that is probably the best.

Such is life.

tanker27 wrote:

FYI, I have had the house wired for 5 AP drops.

Where those runs go back to your switch, you get a power strip to plug the 5 injectors into, 5 shortish patch cables from the switch to the non-powered port of injector, and then plug the pre-existing 5 cable runs to the drops into the powered ports of the injector. Fin.

Make sure you plug the right cable into the right port, and if you are getting a mix of different APs(I did read that you aren't, but just as a general advisory), make sure you don't get the injectors mixed up.

5 APs is a lot for normal North American wood construction. I've got mine mounted on the ceiling of our upstairs landing and it gives good coverage throughout the house and my kids have commented on the WiFi working across the street in a neighbour's house. I can see needing a couple more if your house is longer and lower, but 5 sounds like a lot.

I have a dumb Gigabit switch -> PoE injector -> AP. The switch and injector are down in the wiring closet in the basement and I have conduit running up to the attic and then a hole drilled through the ceiling for the wire in to the AP. The AP comes with a drilling template and mounted to the ceiling looks just like a smoke detector.

notfred wrote:

5 APs is a lot for normal North American wood construction. I've got mine mounted on the ceiling of our upstairs landing and it gives good coverage throughout the house and my kids have commented on the WiFi working across the street in a neighbour's house. I can see needing a couple more if your house is longer and lower, but 5 sounds like a lot.

+1, maybe I just have a smaller house but a single Netgear AP hooked up to an EdgeRouter covers all 3 floors fine. Perhaps the guys with 5 APs have mansions or nuclear bunkers?

The stock router from Comcast doesn't seem to have any issues covering all 3 floors of my house. Granted I did take pains to ensure it is located within a few feet of the physical center of the house.

Well if he built the house or had it renovated, like, having those runs done doesn't really seem like the same level of overkill: better to have and not want, than to want and not have?

Also, that kind of thing can be a selling point in the future: homebuyers routinely pay a massive premium for "newly finished" homes (this is where flippers can make those killings), and this kind of thing in particular can be a tipping point for potential buyer to chose one home versus another.

Just wanted to chime in on the power injectors: I got some UAP‑AC‑PRO and the five pack didn't come with the power injectors. Which was fine by me because I didn't need them. The one I got individually did have the power injector. Check the description to make sure it has them if you need them.

Glorious wrote:

Well if he built the house or had it renovated, like, having those runs done doesn't really seem like the same level of overkill: better to have and not want, than to want and not have?

Also, that kind of thing can be a selling point in the future: homebuyers routinely pay a massive premium for "newly finished" homes (this is where flippers can make those killings), and this kind of thing in particular can be a tipping point for potential buyer to chose one home versus another.

I'm wary of installing any permanent technology in the house. A cable conduit is appealing, but devices are increasingly going wireless. Home security systems used to be a mess of CAT-3 wiring, now it's wireless stuff like Nest/Arlo/Canary that requires a standard outlet. Some are even efficient enough to be solar powered. Running a CAT-6 felt like a great idea until I realized I get 800mbps from a powerline adapter -- that's "good enough" for me. With the shifts to IP-based TV, I bet it's a matter of time before all settop boxes run off of, well, IP infrastructure like your typical wireless AP.

So yeah, low voltage conduit is nice to have, but I bet the next owner may be in an age where the wire requirements for all devices may just be a 120V plug.

He's indicating that he already has them, I was just speculating as to why he might have them, and in such abundance.

I don't have anything like that myself: I have 3 APs none of which are mounted structurally and one of which is linked via powerline networking

Have you verified that you actually get 800mbps? That's an incredible result for laboratory conditions, and in real-world scenarios (it's a half-duplex shared medium and we're in a world beset by noisy switch-mode wall warts etc...) you don't get anything like that.

I have a nice pair in a normal setup, but I average like 100mbps with just iperf testing...

Anyway, I'm not so convinced about inevitability of power networking...

So to answer the burning questions, I had them run during construction when the walls were all open. Obviously, Its way cheaper to run the CAT6 during construction. All the runs terminate to a closet in my office. Yes, its overkill. No, I'm not going to use all of them for APs at least not right now, just two. (2 story, 2535 sq ft house)

Glorious wrote:

Have you verified that you actually get 800mbps? That's an incredible result for laboratory conditions, and in real-world scenarios (it's a half-duplex shared medium and we're in a world beset by noisy switch-mode wall warts etc...) you don't get anything like that.

I have a nice pair in a normal setup, but I average like 100mbps with just iperf testing...

Anyway, I'm not so convinced about inevitability of power networking...

Does it matter? It's good enough. I haven't formally verified it but I was pulling two files down at 40MB/s+ simultaneously. The utility indicates up to 950Mbps down and 400Mbps up, less depending on how noisy the electrics are at any given moment. It's "good enough" for me and saturates a 240mbps link to Netflix and YouTube instantly with about +2ms of delay.

100mbps is pretty dismal, you might try the newer stuff that includes more than just Fast Ethernet. I got that on some AV500 adapters a few years back, but now that I have a place of my own I bought a pair of AV2000s with all the fancy MIMO GBE buzzwords and stuff. My house is younger than me, and the powerlines are separated by a total of perhaps 30 feet of wire across the breaker.

tanker27 wrote:

So to answer the burning questions, I had them run during construction when the walls were all open. Obviously, Its way cheaper to run the CAT6 during construction. All the runs terminate to a closet in my office. Yes, its overkill. No, I'm not going to use all of them for APs at least not right now, just two. (2 story, 2535 sq ft house)

In that case I vote you designate 3 more mancaves and use one AP for each.

tanker27 wrote:

So to answer the burning questions, I had them run during construction when the walls were all open. Obviously, Its way cheaper to run the CAT6 during construction. All the runs terminate to a closet in my office. Yes, its overkill. No, I'm not going to use all of them for APs at least not right now, just two. (2 story, 2535 sq ft house)

That sounds about right!

Duct Tape Dude wrote:

Does it matter? It's good enough. I haven't formally verified it but I was pulling two files down at 40MB/s+ simultaneously. The utility indicates up to 950Mbps down and 400Mbps up, less depending on how noisy the electrics are at any given moment. It's "good enough" for me and saturates a 240mbps link to Netflix and YouTube instantly with about +2ms of delay.

Here's the thing: It's obviously "good enough" for me too, right? After all, I just explicitly said that I use it.

That being said, your results are completely contrary to every single review of these devices that I have ever seen in laboratory conditions. No one gets even close to 600+ mbps. No one. I mean, no one can seem to break 500 by plugging these things literally right next to each other.

Duct Tape Dude wrote:

100mbps is pretty dismal, you might try the newer stuff that includes more than just Fast Ethernet. I got that on some AV500 adapters a few years back, but now that I have a place of my own I bought a pair of AV2000s with all the fancy MIMO GBE buzzwords and stuff. My house is younger than me, and the powerlines are separated by a total of perhaps 30 feet of wire across the breaker.

I have a pair of AV2000s that I bought not even 9 months ago. (PLA5456KIT). I bought them because they were highly rated, because they didn't suffer from insane blufferbloat (the previous AV500 adapters I had given up on ~year+ previous did, grievously so), and because I wanted ~100mbps, which seemed to be best I'd likely be able to achieve realistically (and my estimate matched my reality).

My speed, however "dismal", is something that I formally verify all the time with a standard bandwidth tool (iperf). I am not handwaving or relying on higher-level and notoriously inaccurate/unreliable file system progress indicators.

In other words, I am not making any assumptions here, whereas you are relying on them heavily, even about me in order to (falsely) discredit my experience.

----

Let's say that you do in fact achieve 950 mbps (...which actually requires jumbo frames on normal wired ethernet...), will everyone?

-----

So, why does it matter? Because if your results are extra-ordinary, shouldn't that be factored into your advocacy?

I am not an uninformed person on this subject, I literally use the things and know quite a lot about them, and my (formally verifiable) results were nothing like yours. I think 100 mbps *IS* good enough, but I was drawing attention to how you were claiming performance that is unrealistic.

My experience, as backed up by every serious reviewer of these things, ever, is far, far more common. And I'm also absolutely certain of it.

Going back to the original PoE question...as others have stated, you do need to watch the PoE "standards." A lot of Ubiquiti's equipment is 24V instead of 48V and some PoE switches won't let you change the voltage. However, their AP's are 48V. I'm running mine directly off of an older Extreme Networks X450e-48p without an injector. These switches won't actually try to power a 24V device because it doesn't negotiate correctly. I do intend on upgrading the switch as some point to an X460-48p, which is PoE+.

The one thing that a lot of people neglect to think about with wireless is actual throughput for the number of devices connecting. While you might be able to cover your home with a single AP, that doesn't do much good if that single AP has 20 or so devices connecting to it at the same time. It's already been mentioned in this thread that things are going increasingly wireless so it's not a stretch to have that many devices in a modern, connected home.

I also have CAT6 run throughout my home now. I've opted to wire as many devices as possible.

If you haven't read it, here's a good article from Ars Technica on why Wi-Fi kind of sucks.

curtisb wrote:

The one thing that a lot of people neglect to think about with wireless is actual throughput for the number of devices connecting. While you might be able to cover your home with a single AP, that doesn't do much good if that single AP has 20 or so devices connecting to it at the same time. It's already been mentioned in this thread that things are going increasingly wireless so it's not a stretch to have that many devices in a modern, connected home.

I have 3 APs primarily because 802.11n has 3 non-overlapping (standard/backwards-compatible size) channels.

Glorious wrote:

which actually requires jumbo frames on normal wired ethernet

Errr...no it doesn't.

curtisb wrote:

I also have CAT6 run throughout my home now. I've opted to wire as many devices as possible.

I made this point in regards to powerline networking, in that it is a half-duplex shared medium.

Luckily, that's almost universally deployed as a point-to-point link, at least in DIY/Home applications anyway.

Wifi though? As you point out, not really point-to-point and people have phones, tablets, etc...

So, yes, I wire what I can too.

curtisb wrote:

Errr...no it doesn't.

You can get more than 945mbps on gigabit ethernet with 1500 mtu?

And here I was thinking I wouldn't be pedantic today.

Glorious wrote:

Here's the thing: It's obviously "good enough" for me too, right? After all, I just explicitly said that I use it.

Not obvious enough for stupid me apparently.

Glorious wrote:

That being said, your results are completely contrary to every single review of these devices that I have ever seen in laboratory conditions. No one gets even close to 600+ mbps. No one. I mean, no one can seem to break 500 by plugging these things literally right next to each other.

I don't recommend plugging them in next to one another, might have to start worrying about wave reflections, and also who would seriously do that in real life?

Glorious wrote:

My speed, however "dismal", is something that I formally verify all the time with a standard bandwidth tool (iperf). I am not handwaving or relying on higher-level and notoriously inaccurate/unreliable file system progress indicators.

In other words, I am not making any assumptions here, whereas you are relying on them heavily, even about me in order to (falsely) discredit my experience.

What is protocol here? I am sorry my sad excuse of a homelab doesn't have iperf cron'd every minute in a Dockerized instance across all my LXD VMs atop Proxmox'd bare metal machines (with hourly Ansible deployments and push notifications via Nagios of course) reporting to a Kafka cluster stored into a Hadoop cluster so I can run analytics on exactly when my network slowed down 10%. I was too busy using it for useful things. All hail Glorious, king (I assume?) of all that is known and unknown

I am not worthy
I am not worthy

Glorious wrote:

Let's say that you do in fact achieve 950 mbps (...which actually requires jumbo frames on normal wired ethernet...), will everyone?

Jumbo frames are better for GbE but are really only required for 10GbE. Also I don't get 950mbps at layer 7, that's just what the tool says for layers 2 or 3 probably. I just plugged the thing in and noticed it was stupidfast! And shouldn't you in your unfairly discredited experience note that powerline networking varies by home infrastructure since that's what carries the PHY?

Glorious wrote:

So, why does it matter? Because if your results are extra-ordinary, shouldn't that be factored into your advocacy?

No. Even if my results were ordinary who cares? They're anecdotes just like yours.

Glorious wrote:

I am not an uninformed person on this subject, I literally use the things and know quite a lot about them, and my (formally verifiable) results were nothing like yours. I think 100 mbps *IS* good enough, but I was drawing attention to how you were claiming performance that is unrealistic.

Whoa here we are again, let me pay my Glorious tax, just a moment

Glorious knows all!
We are but mere mortals!
All hail the god that is Glorious and his omniscience and networking omnipotence! May he reign over all with his beloved iperf!

Our Glorious, who knoweth all, hallowed be thy words. Thy righteousness by known, thy expertise be supreme, in iperf as it is in real life.

Glorious wrote:

My experience, as backed up by every serious reviewer of these things, ever, is far, far more common. And I'm also absolutely certain of it.

Wait I thought I just stroked... I need to rest, I can't do this again so soon. But here are some emojis as an IOU:

Using the EdgeRouterX here, along with an older UniFi a/g/b/N AP. Both devices are 24V and powered via the same injector (each shipped with one) using passthrough from router to AP.
Coverage...the AP is installed in a makeshift network closet under the basement stairwell, facing the wrong way from most of the house, and serving two stories of wood-frame above it. In the worst-case second floor bedroom we still get 80% signal strength with no dropouts. Total 1800 square feet covered.

My favorite thing about this setup is the uptime, going on about a year now -- the last time the router went down was when I unplugged everything to intstall a UPS for all the networking gear. Haven't had a reason since then.

As noted by others, take a good look at Ubiquiti's spec sheets for their AC products before deciding how you want to build this, a couple of them use legacy 24V passive PoE and the rest are 48V PoE+.

Look, you are saying that you are doing something that no other product reviewer has even come to close to achieving in *IDEAL* conditions.

Then it turns out that you don't really have any idea how fast the things actually were at all.

I mean, yes, I already knew that because what you were saying was fishy and virtually impossible, but I really get irked when you dismiss my experience (WHICH I MEASURED, NOT HAND-WAVED) by pretending out of whole cloth that my gear is old, I have no idea what I'm talking about, etc...

I mean, come on. I'm trying to provide an honest assessment about reality here, for the benefit of general human knowledge. You're pursuing some sort of ridiculous vanity where you can't just backdown and admit that you overstated something.

Duct Tape Dude wrote:

What is protocol here? I am sorry my sad excuse of a homelab doesn't have iperf cron'd every minute in a Dockerized instance across all my LXD VMs atop Proxmox'd bare metal machines (with hourly Ansible deployments and push notifications via Nagios of course) reporting to a Kafka cluster stored into a Hadoop cluster so I can run analytics on exactly when my network slowed down 10%. I was too busy using it for useful things.

Look, I'm being nice here.

I obviously didn't do anything like that, so why are you goading me?

Iperf is literally two apt-installs and `iperf -c 10.0.0.1` and `iperf -s`.

Done. And then you have a real data point, not hand-waving!

Duct Tape Dude wrote:

I just plugged the thing in and noticed it was stupidfast!

THEN DON'T FALSELY QUANTIFY IT.

Duct Tape Dude wrote:

Also I don't get 950mbps at layer 7, that's just what the tool says for layers 2 or 3 probably.

What tool?

And how does that even make sense?

Look, I said what I did because normals using 1 meter point-to-point wired gigabit CAT-6 don't actually get 950mbps, OK? It was a weird number to see...? Understand?

Duct Tape Dude wrote:

No. Even if my results were ordinary who cares? They're anecdotes just like yours.

No, they aren't. I measured mine. That's data.

You JUST SAID THAT you "noticed it was stupidfast!"

Don't be ridiculous.

Duct Tape Dude wrote:

Whoa here we are again, let me pay my Glorious tax, just a moment

THIS IS A FORUM FOR TECHNOLOGY REVIEW WEBSITE.

I AM TELLING YOU THAT NO OTHER REVIEWER HAS EVER ACHIEVED YOUR EXTRAORDINARY RESULTS.

The only ego being stroked here is yours, because while I am sticking up for OBJECTIVE REALITY and REPRODUCIBLE RESULTS, you are making this about me, DESPERATELY AND INSANELY, because it cannot be about *YOU*

Must be something in the water.

I mean, dude this was my original comment:

Glorious wrote:

Have you verified that you actually get 800mbps? That's an incredible result for laboratory conditions, and in real-world scenarios (it's a half-duplex shared medium and we're in a world beset by noisy switch-mode wall warts etc...) you don't get anything like that.

I have a nice pair in a normal setup, but I average like 100mbps with just iperf testing...

Anyway, I'm not so convinced about inevitability of power networking...

You cannot honestly pretend that I was being even remotely mean, or unpleasant or anything of the sort.

I was very nice and very polite.

You, however, started a meltdown where you called my credibility into question and started this trainwreck because you didn't want to admit that, no, you didn't verify it.

ludi wrote:

Must be something in the water.

Look, I understand this attitude, but I was not trying to start this flame-war.

I am not against powerline networking--- I use it!

But no one, and I mean no one, has managed to achieve results even remotely close to what Duct Tape Dude is (sort of) claiming, even in ideal conditions.

I think that's an important point that should be made when we are talking about solutions for other people who might be reading this thread.

I stand by my initial post in which I wasn't being inflammatory or anything of the sort, I said something that is true and simply asked if he had verified his numbers.

<Mod hat on> OK, that's enough, can we please get back to being nice to each other and discussing network equipment. Stop feeding the trolls or there will be consequences. <Mod hat off>