Indexes

Indexes are essential in improving query performance by enabling fast retrieval of relevant documents when querying large collections.

Indexes in FerretDB are created based on a specific field(s) within documents. Creating an index involves having a data structure that maps the values in the indexed fields to the locations of the related documents, making it possible to retrieve documents more quickly based on those fields.

How to create indexes

Use the createIndexes() command to create an index on a collection. You can use the createIndex() method to call the createIndexes() command.

The createIndexes() command takes two arguments: a document containing the index key (fields to index and direction - either ascending or descending), and an optional document specifying the index options.

You can create single field indexes or compound indexes.

Single field indexes

Suppose a products collection contains the following documents:

{ _id: 1, name: "iPhone 12", category: "smartphone", price: 799 }
{ _id: 2, name: "iPad Pro", category: "tablet", price: 999 }
{ _id: 3, name: "Galaxy S21", category: "smartphone", price: 699 }
{ _id: 4, name: "MacBook Pro", category: "laptop", price: 1299 }

Here's an example of the createIndex method to create an index on the price field of a products collection:

db.products.createIndex({ price: 1 })

This creates an ascending index on the price field.

note

1 specifies the index direction for ascending order. If it's -1, it specifies a descending order direction for the index.

Compound indexes

For compound indexes, you can create an index key combining multiple fields together as a key. Below is an example of a compound index that uses price and category fields from the products collection as the index key:

db.products.createIndex({ price: 1, category: 1 })

Unique indexes

You can create unique indexes to ensure that the indexed fields do not contain duplicate values. To create a unique index, set the unique option as true when calling createIndexes() command.

Below is an example of a unique index for the name field from the products collection:

db.products.createIndex({ name: 1 }, { unique: true })

Unique indexes can be compound. Here is an example of a unique index consisting of the category and name fields from the products collection:

db.products.createIndex({ category: 1, name: 1 }, { unique: true })

Index creation details

• If the createIndexes() command is called for a non-existent collection, it will create the collection and its given indexes.
• If the createIndexes() command is called for a non-existent field, an index for the field is created without creating or adding the field to an existing collection.
• If you attempt to create an index with the same name and key as an existing index, the system will not create a duplicate index. Instead, it will simply return the name and key of the existing index, since duplicate indexes would be redundant and inefficient.
• Meanwhile, any attempt to call createIndexes() command for an existing index using the same name and different key, or different name but the same key will return an error.

How to list indexes

To display a collection's index details, use the listIndexes() command. You can also use the getIndexes() method to call the listIndexes() command.

To return the list of indexes in the products collection, use the following command:

db.products.getIndexes()

The returned indexes should look like this, showing the default index, single field index, and compound index.

{
  cursor: {
    id: Long("0"),
    ns: 'db.products',
    firstBatch: [
      { v: 2, key: { _id: 1 }, name: '_id_' },
      { v: 2, key: { price: 1 }, name: 'price_1' },
      {
        v: 2,
        key: { price: 1, category: 1 },
        name: 'price_1_category_1'
      }
    ]
  },
  ok: 1
}

How to drop indexes

You can also drop all the indexes or a particular index in a specified collection, except the default index (_id).

FerretDB supports the use of the dropIndexes() command. You can also use the dropIndex() method to call the dropIndexes() command to a particular index from a collection.

Using the returned indexes above, let's drop the index with the name price_1.

db.products.dropIndex('price_1')

Another way to perform this action is to use the same index document as the index you want to drop. For the same example above, you can rewrite it as:

db.products.dropIndex({ price: 1 })

Using the dropIndexes() command, specify the index as "*" to remove all indexes from the collection, except the _id index.

db.products.dropIndexes('*')

This will drop all the non-_id indexes from the collection.