On the exact opposite end of the spectrum is #influxdb , whose v2 to v3 migration process is "Export all your data to plaintext and import it into the new DB". No in-place migration is available. What? Seriously? 22GB of timeseries data is going to take a while to round-trip via plaintext.