Question

Suppose you have a few, quite large (100k+) objects in available and can provide this data (e.g. name) in 20+ languages. What is an efficient way to store/handle this data in a SQL database.

The obvious way to do that looks like this – however, are there other ways which make more sense? I’m a bit worried about performance.

  CREATE TABLE "object" (
      "id" serial NOT NULL PRIMARY KEY
  );                                  
  CREATE TABLE "object_name" (
      "object_id" integer NOT NULL REFERENCES "object" ("id")
      "lang" varchar(5) NOT NULL,
      "name" varchar(50) NOT NULL 
  );

As for usage, the use will only select one language and that will result in potentially large joins over the object_name table.

Premature optimization or not, I’m interested in other approaches, if only gain some peace of mind, that the obvious solution isn’t a very stupid one.

To clarify the actual model is way more complicated. That’s just the pattern identified so far.

Answer 1

If you have a combined key on (object_id, lang) there shouldn’t be any joins, just an O(1) lookup, right? (Try with EXPLAIN SELECT to be sure)