Docs Menu

Docs HomeDevelop ApplicationsMongoDB Manual

Queryable Encryption Limitations

On this page

  • Overview
  • MongoDB Support Limitations
  • Contention Factor
  • Manual Metadata Collection Compaction
  • encryptedFieldsMap Modification
  • Read and Write Operation Support
  • Redaction
  • Query Log and Database Profiler Redaction
  • Command and Aggregation Stage Output Redaction
  • Topology Support
  • Sharding
  • CRUD
  • Collections
  • Drop Collection
  • Create Collection
  • Query Types
  • Encrypted Field Names
  • Views
  • Collation
  • Unique Indexes
  • The _id Field
  • Read/Write Query Support

Consider these limitations and restrictions before enabling Queryable Encryption. Some operations are unsupported, and others behave differently.

For compatibility limitations, please read Compatibility.

Enabling Queryable Encryption on a collection redacts fields from some diagnostic commands and omits some operations from the query log. This limits the data available to MongoDB support engineers, especially when analyzing query performance. To measure the impact of operations against encrypted collections, use a third party application performance monitoring tool to collect metrics.

For details, see the Redaction section of this page.

Contention factor is a setting that helps tune performance based on the number of concurrent connections.

You can set the contention factor only when specifying a field for encryption. Once you specify a field for encryption, the contention factor is immutable. If you don't specify the contention factor, it uses the default value of 8.

For more information, see Configuring contention factor.

Manually run metadata collection compaction when your metadata collections exceed 1 GB. Compaction decreases the size of the metadata collections associated with encrypted fields, and improves performance.

Do not modify the encryptedFieldsMap passed to your MongoClient. Doing so can result in unexpected and incorrect behavior when querying.

The mongod only stores encrypted BinData and applies any aggregation expression or query operator specifying an encrypted field against the BinData value. While the expression or operator may support BinData fields, the resulting value may be incorrect or unexpected when compared to issuing that same expression or operator against the decrypted value. The mongod throws an error if the expression or operator does not support BinData values.

To improve security, Queryable Encryption redacts fields from some diagnostic commands on encrypted collections, and omits certain operations from the query log. An encrypted collection is any collection with the encryptedFields option.

The CRUD operations below are omitted from the slow operations query log and the Database Profiler system.profile collection when you run them on an encrypted collection.

The collStats command and $collStats aggregation stage:

  • Omit "queryExecStats"

  • Omit "latencyStats"

  • Redact "WiredTiger", if present, to include only the url field.

The currentOp command and $currentOp aggregation stage:

  • Omit all fields after "command".

  • Redact "command" to include only the first element, $comment, and $db.

The top command only returns the collection name of encrypted collections.

The $planCacheStats aggregation stage omits operations against encrypted collections, even though the operations are cached as normal.

  • Replica sets and sharded clusters are supported

  • Standalone deployments are not supported

  • Secondary reads are not supported

  • Metadata collections must not be sharded

  • Shard keys must not be encrypted fields

  • Queryable Encryption does not support multi-document update operations. db.collection.updateMany() is not supported.

  • Queryable Encryption does not support multi-statement update or delete operations. db.collection.bulkWrite() with more than one update or delete operation is not supported.

  • Queryable Encryption limits db.collection.findAndModify() arguments.

    • fields is not allowed

    • new must be false

  • When performing an upsert operation, any encrypted fields in the filter are excluded from the insert.

  • You cannot use Client-Side Field Level Encryption with Queryable Encryption on the same collection.

    • Queryable Encryption does not support migration from collections encrypted with Client-Side Field Level Encryption. You must decrypt your documents and insert them one-by-one.

    • Queryable Encryption does not support automatic migration from an unencrypted collection. You must insert the documents one-by-one.

  • Queryable Encryption supports new collections only. You cannot add or remove Queryable Encryption to existing collections.

  • You cannot rename a collection with encrypted fields.

  • You cannot specify jsonSchema to db.createCollection() when creating a Queryable Encryption collection if the jsonSchema contains the encrypt keyword.

  • Queryable Encryption does not support Views, Time Series Collections, or capped collection.

  • Queryable Encryption does not support TTL Indexes or Unique Indexes.

  • You cannot disable jsonSchema validation.

  • When a collection uses the encryptedFields option:

    • You cannot set validationLevel to none.

    • You cannot set validationAction to warn.

  • Dropping collections from a MongoClient that is not configured for Queryable Encryption doesn't drop the associated metadata collections. mongod logs a warning if you drop a collection with encrypted fields before dropping the metadata collections.

Tip

When dropping a collection with a MongoClient configured for Queryable Encryption, compatible drivers drop the associated metadata collections as well.

Always explicitly create any collections that use Queryable Encryption. Implicit collection creation doesn't create the necessary indexes and metadata collections, resulting in poor query performance.

The query type you specify for a field when creating the collection in encryptedFieldsMap is immutable. You cannot add new query types to an existing field, and you cannot change the existing query type.

Encrypted field names are immutable. For example, if you specify ssn as a field to encrypt, you cannot $rename the field name to taxPayerID.

Applications cannot rely on the automatic encryption validation to prevent unsupported queries against views on collections with encrypted fields.

Queries against views on collections containing values encrypted with Queryable Encryption may return unexpected or incorrect results if either the underlying view aggregation pipeline or the query references encrypted fields. If creating a view on a collection containing values encrypted with Queryable Encryption, avoid operating on encrypted fields to mitigate the risk of unexpected or incorrect results.

For more information on views, see views.

Queryable Encryption ignores user-specified collations or collection default collations. Encryption obscures the field value and prevents normal collation behavior. Collation-sensitive queries against encrypted fields may return unexpected or incorrect results.

For more information on collations, see Collation Document.

Queryable Encryption-compatible drivers have a list of supported operations for automatic encryption. For unsupported read and write operations, the underlying support library cannot introspect the collection catalog to identify the default collation. Applications therefore cannot rely on the Queryable Encryption validation to prevent querying on encrypted fields with collation defaults.

Unique indexes can't guarantee uniqueness if the index key specifies any encrypted fields.

Queryable Encryption always outputs a different encrypted value, even for the same input. The server considers each encrypted value unique, even though the decrypted value may not be. This means that even if a collection uses an index-enforced unique constraint on fields, the collection can still contain multiple documents with duplicate plaintext values for that field.

Queryable Encryption compatible drivers configured for automatic encryption support some operations for automatic encryption. For unsupported read and write operations, the underlying library can't introspect the index catalog to identify a given field as unique. Applications can't rely on automatic encryption validation to prevent unique constraint violations on randomly-encrypted fields.

You cannot configure Queryable Encryption to encrypt the _id field because it relies on the value auto-generated by MongoDB.

Automatic encryption supports a subset of commands, query operators, update operators, aggregation stages, and aggregation expressions. For the list of supported operations, see Supported Operations for Automatic Encryption.

← Choosing an In-Use Encryption Approach