Is this index redundant? - sql

Question

A recent review of a fairly high traffic table defined as follows:
CREATE TABLE [dbo].[SomeTable](
[Id] [bigint] IDENTITY(1,1) NOT NULL,
[SomeId] [bigint] NOT NULL,
[Time] [time](0) NOT NULL,
[InsertTime] [datetime] NOT NULL,
[SequenceNumber] [int] NOT NULL,
[OtherId] [int] NULL,
CONSTRAINT [PK_Tracks] PRIMARY KEY CLUSTERED
(
[Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]
reveals the following index definitions:
CREATE NONCLUSTERED INDEX [i1] ON [dbo].[SomeTable]
(
[SomeId] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
-and-
CREATE NONCLUSTERED INDEX [i2] ON [dbo].[SomeTable]
(
[SomeId] ASC,
[OtherId] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
This area isn't really my strength, but isn't index i1 superfluous?

Answer 1

Maybe, maybe not. If you were the optimizer, which index would you use for the following query:
select [SomeId]
from [dbo].[SomeTable]
If that query is the kind that's critical to your application and the table is large, having that targeted index could be useful. But you're right in that any query that could be satisfied by i1 could also be satisfied (perhaps more expensively) by i2.

Answer 2

Yes, it is redundant. You'll often end up with this kind of situation when someone adds a new index without reviewing whether it makes any existing indices redundant.
You'll find it worthwhile reading this post that describes circumstances where the apparrently redundant index would be useful. However, since your table doesn't contain large columns, it won't apply to you.
For reference, this blog describes how you can eliminate redundant indices from your databases.

Answer 3

You should not only think about the width of the extra column in the i2 when you are weighting the redundance of the index but also the granularity of it in relation of the first column.
If OtherId has and takes a lot of values for every valor of the SomeId col, queries with only SomeId in the where will take more time without the seemingly redundant index than with it.