I started with Home Assistant on a Raspberry Pi 4 2GB running with an SD card several month ago. Gaining more experience I moved the history/recorder from the integrated database to a MariaDB running on my Synology NAS DS 416j. The NAS itself runs on an ARM and does not support Docker, InfluxDB, or Grafana.
Now I would like speed up the reboot speed of my pi. Therefore I plan to move the installation onto an SSD, also trying to prevent an SD card failure, I luckily did not run into yet.
This is where I come to my question. Is it advised to run any databases (such as MariaDB or InfluxDB) on the Pi’s SSD?
do yourself a favour and don’t put too much money into the Pi. I also switched from a Pi 3 B+ to a mini PC now for several reasons like more power and the possibility to add a normal 2.5" HDD for the log files.
About the logs. With influxDB you will have a whole lot of writes on the ssd. This will decrease the lifetime drastically.
As indicated I started home assistant with the Raspberry Pi as I had no appropriate hardware in my house available to run home assistant. As such the little device is doing a good job. I much appreciate the low power consumption and easy maintenance. With it running Home Assistant OS, there is little to nothing for me to care about beyond updating Home Assistant itself.
I thought maybe with an SSD, I could get more out of the device. I am afraid that the SSD will run into similar problems as the SD cards, however as long as I have been running Home Assistant I luckily did not run into these problems and this is why I am asking for other people’s experience.
If a completely different device is preferred for running home assistant it should be as well low maintenance if possible and low running costs (power consumption).
The only other “always on” devices are my NAS (Synology DS416j) which is purely used as a network attached storage device (okay and the Database…) and a Nvidia Shield TV running the Plex Server. Don’t know if a NUC or anything comparable could replace even all those devices combined and therefore equalize the power consumption.