Is there an advantages/disadvantages (pros/cons) write-up anywhere on HA hardware and installation options?

I’ve been running HA on a Raspberry Pi 3B+ now for about a year and a half and now have over 300 entities with more being added every month. I feel like I could benefit by upgrading my setup but don’t know which route to go and have scoured the internet and this forum to try and find something that lays out all the options in one place and provides the advantages and disadvantages of each so that I can then determine what my best option is. If anyone knows of something like this that exists I’d really appreciate it if you could direct me to it.

In my particular case, I have two options (without going out and buying something new):

  1. An old desktop with a 2.9 GHz quad-core AMD Phenom II X4 840T processor, 8GB of RAM, and a 240GB SSD that I could use as a dedicated HA device or, from my understanding,

  2. I could install HA as a virtual machine (VM) on the new desktop I built last year which has a 3.7GHz 8-core Ryzen 7, 32GB of RAM, and a 2TB SSD which I use for some picture and video editing (no gaming).

I just have no idea which way is the better for my situation, and if there’s a best way to go about setting it all up. Again, I appreciate any help.

IMHO It all depends on what you have around. I run a windows laptop with blueiris so I multi purposed it by also running the hassio OVA as virtualbox guest. Runs great with minimal resources.

I used to run hassio supervised under a ubuntu guest but became a bit messy and overkill since I really didnt need the ubuntu environment. Also there are discussions on retiring that installation method anyhow.

Why don’t you try running hassio ova first. Very easy to do, and if you don’t like it, it’s easy enough to migrate your config to another platform.

To me it’s the best way. If for some reason your guest crashes or gets corrupted, you just need to get a fresh copy of the base ova, do a snapshot restore and you are back in business.

The end-result of that discussion was NOT to retire it. Home Assistant Supervised is an officially supported installation method as described in ADR-0014.

Nor do we. If I told you I have two computers how much can you assume what’s best for my situation? You would have to learn more about my technical knowledge, experience, willingness to learn, and budget to offer solutions tailored to “my situation”.

What expectations do you have that your current system fails to meet? The answer to that may help to guide you (and us) to formulating an appropriate solution.

For me both of your options would probably be too power hungry. Keep in mind that you’ll run this 24/7.

So I guess I would say I know enough to be dangerous :stuck_out_tongue: like I mentioned I built my own desktop, tinker with different types of electronics, etc. I am looking to use my old Raspberry Pi to tasmotize some of my connected devices and am unsure if that would cause any additional strain on my HA setup. Additionally, I have a dozen or so TP-Link smart switches, with many of them setup to refresh every 2 seconds when there’s motion in the room so it can quickly refresh other devices/lights also in the room. I was thinking that having a more robust setup would allow me to reduce that down even further without causing lag.

Power hungry compared to something like a NUC? Is it honestly worth dropping a couple hundred on another device when I have an extra desktop available at no cost? My residential service rate is 5.29 cents a KWH. I figure a desktop running at 90w an hour is really only running me about $3-4 a month. If that’s the case it would take years to recoup the cost of a new device…

If you “tasmotize” devices, their load on Home Assistant is likely to decrease the demand on Home Assistant. Consider how you described the operation of your TP-Link devices which use polling in order to maintain an up-to-date status. Devices running Tasmota rely on MQTT which is event-based. That means the device transmits its changes only when they occur. That means it talks only when it needs to report useful information. Otherwise, it remains silent which means considerably less network traffic (and demand on the receiver, namely Home Assistant) than something operating on a 2-second polling interval. All this to say that through efficiency one can extend the useful life of an RPI3.

Was in a similar situation some years ago and I based on my experience I would go for your first option. Installing HA with the VirtualEnv gives the performance and flexibility that I valued highly, enabling all of my experiments.