Most people don't need 15GB of sensor data, so I don't think this is the best measure of HA being hands-off or suitable for the average person.

What makes you think that this issue is related to the amount of data? I see many reported MariaDB corruption issues after HA upgrade on the forums.

If you need to store 15GB of sensor data, a MySQL derivate is not what I'd choose

15GB is tiny, db selection doesn't really matter at that small of a scale

Depends.

15GB on an actual server? Peanuts.

15GB MariaDB/Mysql database on a Raspberry Pi SD card? Recipe for disaster.

Why is that?

RDBMS isn't the best place to store time series data. There are better and more efficient options for that.

I'm a noob when it comes to DBs, but stumbled upon using InfluxDB with Home Assistant - would you say that's a solid choice, or are there better alternatives out there?

InfluxDB is pretty much the most popular and widely used time series database.

The difference between just storing every value ever and a time series DB is that the latter one can reduce the data frequency when it gets older.

Like if you're measuring your fridge temperature, you might store it every minute. But do you care about 1 minute accuracy when the data is two years old? Would 5 or 15 minutes be enough?

This is what time series DBs do automatically.

They're also optimised for data that's formatted as <time stamp> - <values>, making inserts and queries fast for data like that.

Timeflux or some such are better alternative to store timeseries data.

I have been using HA for 4 years, no such issues. But even if, restore from backup seems like an easy fix?

I think he also had a problem with backups, but I’m not sure.

Almost 9 years, actually.