although I just joined the forum, I’ve been reading and “studying” the wisdom shared here for a couple of years now. In this time I’ve been running a RPi 3 as home server as some sort of “sandbox” to learn and test how OHab works and behave, since right now I’m living on a rented apartment and can’t actually “smarthometize” it.
But recently I bought a house and I will move there once this COVID-19 quarantine ends, so I’m using the free time to design the new network layout, plan the location of the servers and more or less plan the first home automation elements I want to install, so that takes me to the matter of the server.
As most of you would know a really important thing in this whole SmartHome stuff is the WFA (Wife Approval Factor) and besides the aesthetics aspect, this factor also takes into account the reliability of the system, if for some reason it fails “too much”, bye bye smarthome.
Of course, the reliability not only depends on the server, but also in any piece of hardware you set on the whole system (but for the end devices I already make her understand that they can fail like a regular light bulb or a regular doorbell), and I also know that regardless you use a 200€ NUC or a 40€ RPi, the systems can crash anyway, that what the backups and those sort of stuff are ment for.
So back to the point, focusing on the server, I want to reduce to minimum the posibility of failure of the server, but also use persistance and some analytics software (graphana, kibana, or similar…) that’s why I decided to move from a Pi, which is fine for now for testing and experimenting purposes, but will be short for the analytics stuff. That’s why I think on using a NUC.
So my idea is using a very basic NUC, the Celeron based, with one stick of 4Gb of RAM (so I can update to 8 if necessary). And for the drive is where I’m not totally sure.
My first idea was to directly use a sata SSD (the capacity is not that important as with 100Gb will be enough for years) but then you find the problem of the longevity of the SSD using persistance with so many writings on disk (of course the bigger the disk, the more will last since it has more space to loop the writings)
Then my other Idea was to do a mix and use an m.2 unit to run the OS and add an HDD for logs and persistence, but not sure if that’s actually a good idea or if it would slow the system too much.
Then the third option is use my OpenMediaVault NAS as storage for the persistence, but I think that would be overcomplicate the stuff too much, when anyway I was thinking of using the OMV server to backup the persistance location.
So in the end my question is if I should go with only the SSD or with the combination of m.2 drive + HDD.
I know there is a lot of info scattered on the forum, but I would like to do this something more like a poll about the subject.