Understanding MongoDB Wildcard Indexes: Creation, Usage, Behavior, and Limitations

MongoDB 4.2 introduced wildcard indexes, which are useful for collections with dynamic schemas where queries may target any known or unknown field. The guide provides an overview of what wildcard indexes are and the scenarios where they are appropriate.

Example documents and queries

{ "userMetadata" : { "likes" : [ "dogs", "cats" ] } }</code>
<code>{ "userMetadata" : { "dislikes" : "pickles" } }</code>
<code>{ "userMetadata" : { "age" : 45 } }</code>
<code>{ "userMetadata" : "inactive" }

Typical queries might be:

db.colA.find({ "userMetadata.likes" : "dogs" })</code>
<code>db.colA.find({ "userMetadata.dislikes" : "pickles" })</code>
<code>db.colA.find({ "userMetadata.age" : { $gt : 30 } })</code>
<code>db.colA.find({ "userMetadata" : "inactive" })

All of these can be satisfied with a single wildcard index: db.colA.createIndex({ "userMetadata.$**" : 1 }) Creating a wildcard index

Wildcard indexes can only be created on MongoDB version 4.2 or later. To check the feature compatibility version:

db.adminCommand({ getParameter: 1, featureCompatibilityVersion: 1 })

To set it to 4.2:

db.adminCommand({ setFeatureCompatibilityVersion: "4.2" })

Single‑field wildcard index example

{ "_id" : ObjectId("5ee2df16911d8dfaa91520b4"), "product_name" : "Spy Coat", "product_attributes" : { "material" : [ "Tweed", "Wool", "Leather" ], "size" : { "length" : 72, "units" : "inches" } } }</code>
<code>{ "_id" : ObjectId("5ee2df30911d8dfaa91520b5"), "product_name" : "Spy Pen", "product_attributes" : { "colors" : [ "Blue", "Black" ], "secret_feature" : { "name" : "laser", "power" : "1000", "units" : "watts" } } }

Creating an index on a nested field:

db.product_catalog.createIndex({ "product_attributes.$**" : 1 })

This index extracts values from arrays and nested documents, allowing queries such as:

db.product_catalog.find({ "product_attributes.colors" : "Blue" })</code>
<code>db.product_catalog.find({ "product_attributes.secret_feature.name" : "laser" })</code>
<code>db.product_catalog.find({ "product_attributes.size.length" : { $gt : 60 } })

Full‑field wildcard index

To index every field except _id: db.product_catalog.createIndex({ "$**" : 1 }) Wildcard projection can include or exclude specific fields:

db.collection.createIndex({ "$**" : 1 }, { wildcardProjection : { "fieldA" : 1, "fieldB.fieldC" : 1 } })</code>
<code>db.collection.createIndex({ "$**" : 1 }, { wildcardProjection : { "fieldA" : 0, "fieldB.fieldC" : 0 } })

Note: inclusion and exclusion cannot be mixed, except when explicitly including _id.

Behavior based on field type

Object fields: all nested keys are indexed.

Array fields: each element is indexed; if an element is an object, its contents are indexed; multidimensional arrays are indexed as a whole value.

Other types: the scalar value is indexed.

Wildcard indexes iterate through nested objects/arrays until the deepest level, indexing the full path.

Array index positions

Wildcard indexes do not store array indexes, but MongoDB can still use them for queries that include explicit array positions (e.g., parentArray.0.nestedArray.0) up to eight levels; beyond that, a full collection scan may occur.

Limitations

Wildcard indexes are sparse; fields that do not exist are not indexed, so queries with {$exists:false} trigger collection scans.

They cannot directly match an entire object or array using equality or $ne operators.

Supported index types include Compound, TTL, Text, 2d, 2dsphere, Hashed, and Unique.

Queries using $ne null on arrays or other fields will not use a wildcard index.

Conclusion

Wildcard indexes are helpful for early‑stage schema design where exact field indexes may be missed, but relying on them for all precise queries leads to performance degradation compared to targeted indexes as data grows; therefore, they should complement rather than replace conventional indexes.