Question

I know this is very basic and given where I’m at with Databases I should probably understand this by now, but I do not fully comprehend what an index really is.

To make it concrete, the following is the T-SQL generated by an ORM – it creates an FK and then creates an index for it.

What does having the index do (in the posative sense) and what does excluding it do (in the negative sense)?

-- Creating foreign key on [item_id] in table 'CakeStats'
ALTER TABLE [dbo].[CakeStats]
ADD CONSTRAINT [FK_CakeStat_Item]
    FOREIGN KEY ([item_id])
    REFERENCES [dbo].[Items]
        ([id])
    ON DELETE NO ACTION ON UPDATE NO ACTION;

-- Creating non-clustered index for FOREIGN KEY 'FK_CakeStat_Item'
CREATE INDEX [IX_FK_CakeStat_Item]
ON [dbo].[CakeStats]
    ([item_id]);
GO

Answer 1

An index is a balanced tree that allows faster searching for specific data. Indexing something in a general sense is marking it for easy finding. Visualize an index in the back of a reference book. It’s alphabetically sorted to allow for quick finding of where the actual information is at the cost of extra pages in the back of the book making it larger/thicker. The more stuff you index, the larger that is. There is also a difference between a clustered and nonclustered index. The example I gave above is a nonclustered index so it’s in addition to the normal book contents. A clustered index is the actual content itself sorted by the keys of the index.

In the foreign key case, if you foreign key something, the way the key is enforced is by doing a check on the referenced table/column to see if it exists. Having an index on the foreign key greatly speeds up the look up (log(n) time) rather than require looking at every single row one at a time (n time).

In your example, the index only helps if a query searches CakeStats based on item_id. It actually is very important that Items(id) also has an index or the check on foreign key is going to be expensive every time item_id is inserted/deleted/updated in CakeStats.