Remove Unnecessary Indexes
Overview
Indexes support the efficient execution of queries in MongoDB. However, each index you create has a negative performance impact on writes and requires some disk space.
Creating unnecessary indexes leads to a bloated collection and slow writes. Consider each query your application performs and whether it justifies an index. Remove indexes that are unused, either because the field is not used to query the database or because the index is redundant.
You should identify and drop unnecessary indexes. Use the Atlas UI to manage indexes in Atlas.
Example
Consider a game that awards coins
to players. When a player reaches
20 coins
, that player earns 1 star
and their coins
are
reset to 0. The game has a players
collection with documents such
as the following:
// players collection { "_id": "ObjectId(123)", "first_name": "John", "last_name": "Doe", "coins": 11, "stars": 2 }
The players
collection has an index for every field:
_id
is indexed by default.{ last_name: 1 }
{ last_name: 1, first_name: 1 }
{ coins: -1 }
{ stars: -1 }
last_name
Index
In this example, when the game queries the database for player
information, it finds a single record using a player's full name. The
compound index { last_name: 1,
first_name: 1 }
covers this case, so the game should drop the index
{ last_name: 1 }
because it is redundant.
coins
Index
In this example, the coins
field is never used to search the
database. The game should drop the index { coins: -1 }
because it
is unused.
stars
Index
In this example, at the end of a game, the players' names are displayed
on a leaderboard in descending order by number of stars. The game
should maintain the index { stars: -1 }
, even if it is used
infrequently, to avoid scanning every document in the
players
collection.
Now, the game uses the following indexes:
_id
is indexed by default.{ last_name: 1, first_name: 1 }
{ stars: -1 }
After dropping unnecessary indexes, the players
collection has
more free space and can perform faster writes. The most frequent reads
do not experience a drop in performance because the indexes supporting
those reads still exist in the collection.
Identify & Drop Unnecessary Indexes
To identify unnecessary indexes in Atlas, View Indexes with the Atlas UI.
Under the Indexes tab in Data Explorer, you can view Size, Usage, and other information for each of your indexes. If an index is unused or covered by another index, you should drop it.
To drop an index in Atlas, Drop an Index with the Atlas UI.
Learn More
To learn more about indexes, see Indexes.
To learn more about indexing for your use case, see Indexing Strategies.
MongoDB also offers a free MongoDB University Course on improving database performance, including optimizing indexes: Monitoring and Insights.