3 ms·
these days if you are modeling the "sparse, high-dimensional data" of the kind that EAV schema does well, you probably should be reaching for JSON columns inste
by paulmd 5y ago
these days if you are modeling the "sparse, high-dimensional data" of the kind that EAV schema does well, you probably should be reaching for JSON columns instead of EAV schema.
- jiggawatts 5y agoSome databases support "sparse columns", which use EAV under the hood but are better optimised: https://docs.microsoft.com/en-us/sql/relational-databases/tables/use-sparse-columns https://docs.microsoft.com/en-us/sql/relational-databases/ta...
- Serow225 5y agoThanks, I forgot about that! :)
- smitty1e 5y agoIsn't that the problem that these "columnar store" DBs address? A bunch of dictionaries containing UUID/data, where each dimension is its own dict, and the dicts might or might not be running on the same system?