11. Indexes and performance

Indexes dramatically improve query performance.

What is an index?

An index is a special data structure that stores a small portion of the data in a way that is easy to traverse. It is like the index of a book.

Without an index, MongoDB must scan the entire collection (collection scan). With an index, MongoDB can go directly to the relevant documents.

Viewing indexes

To see the indexes of a collection:

db.unicorns.getIndexes()

By default, all collections have an index on _id.

Creating indexes

To create a simple index:

// Index on the name field
db.unicorns.createIndex({ name: 1 })

• 1 = ascending order
• -1 = descending order

For simple indexes, the order doesn't matter. For compound indexes, it does.

Compound indexes

You can create indexes on multiple fields:

db.unicorns.createIndex({ gender: 1, vampires: -1 })

This index is useful for queries that filter by gender and/or sort by vampires.

Unique indexes

To guarantee there are no duplicates:

db.users.createIndex({ email: 1 }, { unique: true })

Now you won't be able to insert two users with the same email.

Removing indexes

To remove an index:

db.unicorns.dropIndex({ name: 1 })

To remove all indexes except _id:

db.unicorns.dropIndexes()

Analyzing queries with explain

To see whether a query uses indexes:

db.unicorns.explain().find({ name: 'Aurora' })

Look for the winningPlan field:

• COLLSCAN = full scan (bad, slow)
• IXSCAN = uses index (good, fast)

Indexes on arrays

You can create indexes on fields that are arrays:

db.unicorns.createIndex({ loves: 1 })

This creates a multikey index that indexes each value in the array.

Indexes on embedded fields

You can create indexes on fields of embedded documents:

db.users.createIndex({ 'address.city': 1 })

Note the dot notation in quotes.

When to create indexes

Create indexes when:

• Slow queries: if a query takes too long
• Search fields: if you frequently filter by a field
• Sorting: if you frequently sort by a field

Don't create unnecessary indexes:

• They slow down writes (INSERT, UPDATE, DELETE)
• They take up disk space
• Each index consumes memory

Collection statistics

To see statistics for a collection:

db.unicorns.stats()

You will see information about:

• Number of documents
• Size of the data
• Size of the indexes

Solutions