I think Home Assistant is a brilliant, low-budget, learning platform. That’s why I use it
Seriously, however you get get it setup and running (VM, Docker, etc.) will be ok as long as you know the limitations of each installation type. And you can change to a different installation type later if you need.
Best of luck and, speaking on behalf of others in this community, we’re looking forward to helping you.
P.S. backups are your friend. Don’t just do them, test that you can recover using them.
Another issue that you need to address might be what types of devices: Z-Wave, Zigbee & Wireless? Each has advantages and disadvantages. If you don’t have wireless everywhere, the extra range of Z-Wave in a large building(s) might be a big advantage. Matter is still a small player, but many expect it to be ‘the future’.
HA leaves commercial systems to dust, assuming the technical skill is there to support the platform. From someone with commercial and residential experience here are the pointers:
The HA platform of choice should be the least demanding in terms of admin overhead. To this end you could consider:
x86 HW architecture, i5 or i7 based or AMD Ryzen based. A good option is the MinisForum MS-01. I suggest barebone then add 32GB RAM and at least 2 x 2TB for redundancy. Plan the storage capacity carefully, especially if you plan adding CCTV in the future, as you want to avoid having to redo the platform because you run out of digital storage.
A virtualised environment, I simply suggest Proxmox as the most desirable option.
HAOS running in a VM.
With the exception of a handful of add-ons most should run outside HA as KVMs/LXCs under the virtualised platform
Regardless of the comments you might have seen, Tuya is a good budget option to start. With Tuya the secret lies in the reliability of the HW manufacturer; you will see hundreds of manufacturers providing sometimes identical HW packaging, but not all of them are created equal
For lights you want Zigbee as it will provide similar experience to normal light switching; however if you have good WiFi, it will also be on a par with Zigbee and can scale better/easier; WiFi has a higher maintenance cost, especially setting up, but provides greater flexibility from all aspects; avoid Zwave at any time, and Matter at the moment.
Always remember, as an admin you need to apply the KISS principle, wherever possible.
If you are open to suggestions I would like to support you for free outside the realms of this forum. I might also be able to assist you with free HW. In any case feel free to reach out if you wish.
The most important question you need to answer is: what is the purpose of implementing an IoT solution; productivity increase, cost savings, safety, etc.
Finally, planning prior to methodically executing is always highly desirable. Make a list of the control & automation realms you want to address first: e.g. lighting, HVAC, covers, electrical, mechanical, plumbing, security, sensors, actuators, etc.
Start with a small stand-alone server, like a Raspberry, a NUC or even an old laptop or desktop.
Make your first installation with some of the devices and automations set up and then go from there.
If sounds like it might be somewhat critical once up and running, so backup is important.
You should at least backup locally and offsite, but if you run in a VM, then you might also back up the VM, which should also be offsite too.
And make sure that the school understand that it might require some technical knowledge to run update and maintain the system, which makes you a key employee and maybe harder to replace in the event something happens to you.
Avoid WiFi!
You are a small school, which probably mean you have plenty of devices that use WiFi and adding more might overwhelm your access points, if they are not industrial grade. Adding more access points is not an easy task, since it requires a good knowledge about frequency measuring and planning, unless you have industrial grade access point controllers, but these are also expensive.
Zigbee might seem like the quick choice, but they use the same frequency as your WiFi and if you have multiple access points, then it is hard to find a good frequency, especially if your access points are not using fixed frequencies.
Z-Wave devices might be a bit more expensive, but they use a lower frequency that have better penetration and are much less used. The reason they are a little more expensive is also that it is a certification required protocol, so you know that devices are tested and made sure that they just works.
Zigbee is an open protocol and might require all kind of jumps to make devices work, like install quirks, avoiding certain brands as routers (the name for a repeater in the Zigbee network) and so on.
The first thing this person need to do is think and/or write on piece of paper his/her goals and objectives. This can take from a few mins to a few hours and it does not have to be complicated. Just a clear understanding of the motivation behind what s/he is trying to accomplish.
Then s/he will need help to see if the infrastructure in the school can meet/accomplish the goals/objectives.
As for Zwave it is an American focused technology. The person is in Australia which follows EU regulations as most of the world does.
A simple rule of thumb is if the Chinese support the technology then it is a safe long term bet. Chinese have thrown their weight behind Zigbee and other mostly open technologies.
I have no idea how available Z-Wave are around the world and yes Z-wave is a lake and Zigbee is an ocean, true.
But Zigbee is also 2.4Ghz (unless you use Zigbee Pro, which is just a puddle in this terminology) and I know something about WiFi on schools, which is not a good combination.
Thanks for highlighting this. We have good1 WiFi accross all four buildings (3 floors each) but there are some places there isn’t great signal. However, the issue may be that this is surrounded by other buildings with businesses who think putting their WiFi on maximum power gives them the best results, so the 2.4GHz spectrum is extremely congested. With a lot of IoT devices being 2.4GHz, I should probably be cautious.
How well does Zigbee and Z-Wave co-exist within congested 2.4GHz?
1 Good WiFi is actually UniFi which isn’t great, but it is at the right price point.
We have Hyper-V infrastructure at the moment and I do have enough linux skills to get myself into trouble. I’d likely put it in there because our backup infrastructure (Veeam) takes full VM snapshots and can restore quickly. Do you see any potential issues with that?
We have four buildings of three floors which are quite close together. If we went ZigBee, do you think it would be better to have one zigbee unit per building (eg: TubesZB PoE) or rely on hardwired devices which can be repeaters?
Thank you, that is very generous. I’ll keep it in mind once I have a better plan in place on how to approach all this.
One issue which I have observed in my own experimentation is that Australia seems to be quite the special snowflake. Most “smart” HVAC controllers I see referred to in the community won’t work in Australia. Smart plugs which work local only are hit and miss. (so probably make my own using Shelly units in a box or something)
Thanks, that is sage advice. Across the four bulidings of 3 floors each, we have 25 APs. They’re UniFi, but prior to this job I sold, deployed, and managed UniFi networks as part of my role as a level 3 with a MSP. The coverage is reasonable, but can get sketchy in places. My first risk is that we’re surrounded by other buildings with businesses running their own networks blasting out on 2.4GHz at max power. Right now at 8pm local I’m seeing some APs with 74% 2.4GHz channel utilisation. (and yes, all the channels are correctly set and only on 20MHz width)
The other big risk is that UniFi loves calling their betas “General Release” and getting unsuspecting users to test them. I’ve seen the weirdest stuff just stop working over the last 10 years I’ve used them. If I use any WiFi devices, I’ll go gently and test extensively before deploying more.
You use Hyper-V as a type-2 or type-1 hypervisor. If you use it as type-2 you are in for long term trouble.
Zigbee coordinators have limitations how many devices you can connect. For example, wired are between 100-300, WiFi 50-100. For simplicity I would use PoE Zigbee coordinators (I highly recommend ZigStar UZG-01) per building and if range is an issue you can use WiFi Zigbee. It depends how many routers/repeaters you can have between battery powered devices and coordinator.
If you use chilled-water FCU you can disregard what the community is saying. The best way for commercial buildings is not to use traditional on-wall thermostats. I would fully automate HVAC and throw controls away from user hands. You can provide temperature control but even that will not be the most efficient way to run HVAC.
Zigbee can coexist with WiFi very well. If you have capable WAPs there should be no issues. A capable WAP is one which can handle congestion well. A lot of people, especially in the USA, consider Unify as a viable option. It is not. Their HW is only basic and does not scale well.
Personally, whenever I encounter Unify I provide two options:
Zigbee and WiFi use pretty much the same bandwidth and with 25 APs you will be using the entire WiFi spectrum and also with heavy loads when suddenly your Os on all machines gets a big update or similar.
Here is a picture of the coexistence of the Zigbee and WiFi 2.4Ghz channels.
Z-Wave is using 800/900 Mhz frequency bands, which are way less congested, because heavy bandwidth usage are not allowed there.
800/900Mhz also have better penetration than 2.4Ghz.
Zigbee and WiFi do not use the same bandwidth. They share the same RF spectrum split in 5MHz bands. Zigbee has a max 10m distance and does not constantly chat, plus its packet size is tiny.
If your WAP supports 200+ clients you are good to. But if you have the lesser varieties which can handle < 100 clients and do not have optimised algorithms to handle congestion, you might want to consider your options.
Not everything is created equal and if you know what you are doing the sky is the limit. Plus you might opt to localise Zigbee in classrooms, not in common areas.
Another option is RF in classrooms with WiFi gateway for every classroom. This option might be the most cost effective for the school.
One more thing to throw at you. You might also want to consider Node-RED to run alongside HA, in a separate VM. Some automations might be more intuitive than HA as it uses visual flow based tooling.
I’m using HA for something similar, but rather than starting from scratch I’m moving over from a punch of separate set top box\cloud type systems that I want to bring in house.
On one hand HA does pretty much everything that I need, but on the other it’s a little bit of a transition moving from companies with dedicated support services to what is essentially a community based system. Getting support can be hit and miss, mostly because it’s “help” rather than “support” as people will often remind you. Sometimes it’s fantastic and someone will walk you through a problem from start to finish, other times it make me long for the teen on the desk in Target. You’re millage may vary.
I think that the important question here is less about how you want to host your system and more what is it that you actually want to do with it?
For example are you looking for a simple system to switch your lights on and off, or do you want to make it into a security system, or something completely different.
Do you have any existing hardware such as sensors, CCTV camera or smart thermostats, or are you looking to buy new equipment in, and if it’s the latter do you have any outlets that stock the right hardware. I’m mostly using legacy systems, but much of the new equipment I’ve had to import.
Of note, when looking for for HA compatible zigbee devices, not everything that’s zibgee actually calls itself zigbee. Sometimes they use a brandname like Hue, or they say that they’re “compatible with Hue”. Most of my Zigbee lighting said that it was ZLL 3. Which is a whole other thing to get into.
I’d likely put it in there because our backup infrastructure (Veeam) takes full VM snapshots and can restore quickly. Do you see any potential issues with that?