JSONB search in Postgresql?

Question

PostgreSQL

The administrator creates fields by directly specifying their type string, int, etc. (this is a separate Fields table), but the user must fill in the same fields with data, so the UserData table is needed, in which data can be presented in two versions:

id:int | value:string | field_id:int

or

id:int | data:jsonb

where the key can be field ID
Fields
| id | type |
| 1 | integer |
| 2 | varchar |
| 3 | real |
user data
| data |
| {1:23, 2:"test string", 3:45.56} |
That is,
in the first case, each record contains string data and refers to a specific field,
in the second case: data for all filled fields is contained in one record.
Which method is preferable if you need to take into account the speed of data sampling for a particular field?
The value for this field can be integer or real, which does not allow for correct sorting in the first method, more precisely, it will be sorted as a string

Answer 1

If you understand your task correctly, selecting from one table on a JSONB field indexed by GIN using the @> operator will be much faster than selecting from several related tables (EAV model).
Proof
Read more here in the comments to the answers