Hi trying to test the questdb for our needs. Generated test data for a hour and measured used disk space. What I find suspectful - constant, incrementing and random values are resulting in exact same disk size - 867 MB, compared to 19, 226, 261 MB in case of influxdb it also looks as a pretty bad compression What could I do wrong?

maybe nothing. What are the table column types?

uuid, string, timestamp in ns, and float value

is string length random? Or is it tag / symbol? How many rows you generated?

uuid and string are symbols, 10000 series of them generated one row per second for a hour 10000 series x 3600 s = 36_000_000 rows

Should be 20 bytes per record.

so there is no gorilla or lz4 compression?

No

ok, the numbers are looking realistically then thank you

hey @vilinski it is possible to use zfs compression (compresses 3x roughly), @Andrey Pechkurov will be able to share some specifics about that

Hi, You could try using ZFS with enabled lz4 compression. In our tests it provides around 2.5-3x compression ratio

@vilinski just to make sure you see this! thanks

👍