Nonclustered indexes in RDBMS can be larger than the tables. It’s usually poor design or indexing a very simple schema in a non-trivial way, but the ultimate goal of the index is speed, not size. As long as you can select and use only a subset of the index based on its ordering it’s still a win.