Explain Atlas Vector Search Results

On this page

Syntax
Verbosity
Explain Results
collectors
allCollectorStats
metadata
query
args
stats
resourceUsage
Examples
allPlansExecution
queryPlanner
executionStats

When you run an Atlas Vector Search query with the explain method, the query returns a BSON document containing query plan and execution statistics that describe how the query was run internally.

Syntax

db.<myCollection>.explain("<verbosity>").aggregate([
   {
     "$vectorSearch": {
       "exact": true | false,
       "filter": {<filter-specification>},
       "index": "<index-name>",
       "limit": <number-of-results>,
       "numCandidates": <number-of-candidates>,
       "path": "<field-to-search>",
       "queryVector": [<array-of-numbers>]
     }
   }
])

Verbosity

Verbosity mode controls the behavior of explain and the amount of information returned. Value can be one of the following, in order of decreasing verbosity:

allPlansExecution	Information about the query plan including the `stats` fields, which contain execution statistics for the query. Includes partial execution data captured during plan selection.
executionStats	Information about the query plan including the `stats` fields, which contain execution statistics for the query.
queryPlanner (default)	Information about the query plan. Does not include the `stats` fields, which contain execution statistics for the query.

Explain Results

The explain method returns a BSON document with the following fields.

Option	Type	Necessity	Purpose
`collectors`	document	Optional	Describes the execution statistics for the collector.
`metadata`	document	Optional	Contains useful metadata.
`query`	document	Optional	Describes the execution statistics for the query. This is not present at the the top level for `indexPartitionExplain`.
`resultMaterialization`	document	Optional	Details related to retrieving per-document data after query execution. This is not returned for `queryPlanner` verbosity mode.
`resourceUsage`	document	Optional	Details the resource usage of the query when it was executed. This is not returned for `queryPlanner` verbosity mode.

`collectors`

The collectors is a BSON document with the following field:

Field	Type	Necessity	Purpose
`allCollectorStats`	document	Required	Statistics of all collectors of the query. Statistics reported represent either the maximum value across all collectors used in the query or a sum of the statistic across all the sub-collectors. The timing statistics are summed to reflect the total time spent across all collectors for the entire query. To learn more, see `allCollectorStats`.

`allCollectorStats`

The allCollectorStats is a BSON document that describes collector statistics across all collectors specified in the query. It contains the following keys:

Field	Description
`collect`	Tracks the duration and number of results collected by the collector.
`competitiveIterator`	Statistics tracking the total duration and the number of times a `competitiveIterator` was requested from the collector.
`setScorer`	Statistics tracking the total duration and number of times a scorer was set on the collector.

`metadata`

The metadata contains helpful metadata, such as the following:

Field	Type	Necessity	Purpose
`mongotVersion`	String	Optional	Current version of `mongot`.
`mongotHostName`	String	Optional	Human readable label that identifies the `mongot` host.
`indexName`	String	Optional	Atlas Vector Search index used in the query.
`cursorOptions`	Document	Optional	Cursor options given to `mongot`.
`totalLuceneDocs`	Integer	Optional	Total number of documents in the index including deleted documents.

`query`

The explain response is a BSON document with keys and values describing the execution statistics for the query. The explain document in the result set contains the following fields:

Field	Type	Necessity	Purpose
`path`	string	Optional	Path to the queried embedding field, only if it isn't the root.
`type`	string	Required	Name of the type of vector search query. See `query` for more information.
`args`	document	Required	Vector search query information. See `query` for more information.
`stats`	document	Optional	`stats` for the query if `explain` ran with `executionStats` or `allPlansExecution` verbosity.

`args`

The explain response contains information how the query was executed internally. The args field includes the following details:

Query type
A structured summary example for each query type
Query options in the structured summary

The following sections describe the query types and the fields in the structured summaries:

WrappedKnnQuery

A wrapper query used for ANN vector search that combines multiple queries. The structured summary includes details on the following option:

Field	Type	Necessity	Description
`query`	Array	Required	Array of sub-queries used in the vector search, typically containing `KnnFloatVectorQuery` and `DocAndScoreQuery`.

KnnFloatVectorQuery

A query type specific to ANN search on float vectors. The structured summary includes details on the following options:

Field	Type	Necessity	Description
`field`	String	Required	The vector field path being searched.
`k`	Integer	Required	Number of nearest neighbors to retrieve.

DocAndScoreQuery: A query type that handles document matching and scoring. The structured summary typically includes statistics about query execution.

ExactVectorSearchQuery

A query type for ENN vector searches. The structured summary includes details on the following options:

Field	Type	Necessity	Description
`field`	String	Required	The vector field path being searched.
`similarityFunction`	String	Required	The similarity function used (`dotProduct`, `cosine`, or `euclidean`).
`filter`	Document	Optional	Pre-filter query that limits the vector search scope.

BooleanQuery: If you use a pre-filter in your vector search query, your explain results include the BooleanQuery type. To learn more about the fields for this type and other query types specific to your pre-filter query, refer to the Query Types on the Atlas Search Explain Page.

DefaultQuery

Queries that are not explicitly defined by another query are serialized using the default query. The structured summary includes details on the following option:

Field	Type	Necessity	Description
`queryType`	String	Required	Type of query.

`stats`

The explain response for executionStats and allPlansExecution verbosity modes includes a stats field that contains information on how much time a query spends in various stages of query execution.

Timing Breakdown

The timing breakdown describes execution statistics pertinent to an areas of query execution. The following fields show the timing breakdown:

Field	Type	Description
`millisElapsed`	Long	Approximate wall-clock time elapsed performing tasks in this area including the amount of time the children of the query spent in this area. The value is approximate number of milliseconds elapsed while performing tasks in this area.
`invocationCounts`	Document	Number of invocations of tasks included in this area. The value is a map of task names to their invocation count.

Areas of Query

Statistics are available for the following areas of query:

Option

Description

context

Statistics related to the execution of the vector search query. There are two tasks whose invocation counts are enumerated in this area:

`createScorer`	Scorer iterates over documents and generates a score for each document. Invocations of `createScorer` create the object responsible for scoring. Note that time associated with this task is not time spent actually scoring documents. Count includes the number of `scorerSupplier` invocations.
`createWeight`	Weight stores state associated with a query and `IndexSearcher`. Count includes the number of `createWeight` invocations.

The time spent in this area is related to the structure of the query, and is not based on the number of results that are iterated through and scored.

For example:

"context" : {
  "millisElapsed" : NumberDouble(4.934751),
  "invocationCounts" : {
    "createWeight" : NumberLong(1),
    "createScorer" : NumberLong(10)
  }
}

match

Statistics related to iterating over and matching result documents. This statistic shows the time it takes to determine which document is the next match. Time spent matching results can vary significantly depending on the nature of the query. There are two tasks whose invocation counts are enumerated in this area:

`nextDoc`	Requests to advance to the next document of the result set. This involves identifying and moving past skips, or other tasks necessary to find the next match. Count includes the number of `nextDoc` and `advance` invocations.
`refineRoughMatch`	Performs a more thorough match. Some queries execute in a two-phase process where a document is first "roughly" matched, and is checked with a second, more thorough phase only after satisfying the first rough match. The `refineRoughMatch` task is the second phase of the two-phase process. Count includes the number of `refineRoughMatch` invocations.

For example:

"match" : {
  "millisElapsed" : NumberDouble(4.901597),
  "invocationCounts" : {
    "nextDoc" : NumberLong(541),
    "refineRoughMatch" : NumberLong(0)
  }
}

score

Statistics related to scoring documents in the result set. There are two tasks whose invocation counts are enumerated in this area:

`score`	Scores each document in the result set. Count includes the number of `score` invocations.
`setMinCompetitiveScore`	Ignores documents whose score is less than the given value. Indicates that a query may have been able to reduce the number of scoring operations performed by ignoring documents with scores below some uncompetitive threshold. Count includes the number of `setMinCompetitiveScore` invocations.

For example:

"score" : {
  "millisElapsed" : NumberDouble(3.931312),
  "invocationCounts" : {
    "score" : NumberLong(536),
    "setMinCompetitiveScore" : NumberLong(0)
  }
}

`resourceUsage`

The resourceUsage document shows the resource used for running the query. It contains the following fields:

Field	Type	Necessity	Purpose
`majorFaults`	Long	Required	Number of major page faults, which occur when the system can't find the required data in memory resulting in reading from the backing store such as disk, during query execution.
`minorFaults`	Long	Required	Number of minor page faults, which occur when the data is in the page cache, but hasn't yet been mapped to the process' page table.
`userTimeMs`	Long	Required	Amount of CPU time, in milliseconds, spent in user space.
`systemTimeMs`	Long	Required	Amount of CPU time, in milliseconds, spent in system space.
`maxReportingThreads`	Integer	Required	Maximum number of threads that `mongot` used during query execution across all batches. For non-concurrent explain queries, the value is `1`.
`numBatches`	Integer	Required	Total number of batches that `mongot` was requested when processing the query.

Examples

The following examples use the explain method on sample ANN and ENN queries. If you want to run the following examples in your own environment, you must first complete the Atlas Vector Search Quick Start using mongosh.

`allPlansExecution`

The following example runs the explain method on an ANN and ENN vector search query for the plot_embedding field with the allPlansExecution verbosity mode.

db.embedded_movies.explain("allPlansExecution").aggregate([
  {
    "$vectorSearch": {
      "index": "vector_index",
      "path": "plot_embedding",
      "queryVector": QUERY_EMBEDDING,
      "numCandidates": 150,
      "limit": 10
    }
  }
])

{
  explainVersion: '1',
  stages: [
    {
      '$vectorSearch': {
        index: 'vector_index',
        path: 'plot_embedding',
        queryVector: [
            -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
            ...
            -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
        ],
        numCandidates: 150,
        limit: 10,
        explain: {
          query: {
            type: 'WrappedKnnQuery',
            args: {
              query: [
                {
                  type: 'KnnFloatVectorQuery',
                  args: { field: '$type:knnVector/plot_embedding', k: 150 },
                  stats: {
                    context: { millisElapsed: 0 },
                    match: { millisElapsed: 0 },
                    score: { millisElapsed: 0 }
                  }
                },
                {
                  type: 'DocAndScoreQuery',
                  args: {},
                  stats: {
                    context: {
                      millisElapsed: 0.011281,
                      invocationCounts: {
                        createWeight: Long('1'),
                        createScorer: Long('2')
                      }
                    },
                    match: {
                      millisElapsed: 0.045831,
                      invocationCounts: { nextDoc: Long('151') }
                    },
                    score: {
                      millisElapsed: 0.049853,
                      invocationCounts: {
                        score: Long('150'),
                        setMinCompetitiveScore: Long('30')
                      }
                    }
                  }
                }
              ]
            },
            stats: {
              context: {
                millisElapsed: 3.226133,
                invocationCounts: {
                  vectorExecution: Long('1'),
                  createWeight: Long('1'),
                  createScorer: Long('2')
                }
              },
              match: {
                millisElapsed: 0.045831,
                invocationCounts: { nextDoc: Long('151') }
              },
              score: {
                millisElapsed: 0.049853,
                invocationCounts: {
                  score: Long('150'),
                  setMinCompetitiveScore: Long('30')
                }
              }
            }
          },
          collectors: {
            allCollectorStats: {
              millisElapsed: 0.211472,
              invocationCounts: {
                collect: Long('150'),
                competitiveIterator: Long('1'),
                setScorer: Long('1')
              }
            },
            facet: { collectorStats: { millisElapsed: 0 } }
          },
          metadata: {
            mongotVersion: '1.44.0',
            mongotHostName: '<hostname>.mongodb.net',
            indexName: 'vector_index',
            totalLuceneDocs: 3483
          },
          resourceUsage: {
            majorFaults: Long('0'),
            minorFaults: Long('0'),
            userTimeMs: Long('0'),
            systemTimeMs: Long('0'),
            maxReportingThreads: 1,
            numBatches: 1
          }
        }
      },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('29')
    },
    {
      '$_internalSearchIdLookup': { limit: Long('10') },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('29')
    }
  ],
  queryShapeHash: '9EFFBE6AC5B8DC66D20DB8939CF3ED6669D0C09BB6938C17AD3FB8C7793765B8',
  serverInfo: {
    host: '<hostname>.mongodb.net',
    port: 27017,
    version: '8.0.5',
    gitVersion: 'cb9e2e5e552ee39dea1e39d7859336456d0c9820'
  },
  serverParameters: {
    internalQueryFacetBufferSizeBytes: 104857600,
    internalQueryFacetMaxOutputDocSizeBytes: 104857600,
    internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
    internalDocumentSourceGroupMaxMemoryBytes: 104857600,
    internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
    internalQueryProhibitBlockingMergeOnMongoS: 0,
    internalQueryMaxAddToSetBytes: 104857600,
    internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
    internalQueryFrameworkControl: 'trySbeRestricted',
    internalQueryPlannerIgnoreIndexWithCollationForRegex: 1
  },
  command: {
    aggregate: 'embedded_movies',
    pipeline: [
      {
        '$vectorSearch': {
          index: 'vector_index',
          path: 'plot_embedding',
          queryVector: [
              -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
              ...
              -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
          ],
          numCandidates: 150,
          limit: 10
        }
      }
    ],
    cursor: {},
    '$db': 'sample_mflix'
  },
  ok: 1,
  '$clusterTime': {
    clusterTime: Timestamp({ t: 1740781674, i: 1 }),
    signature: {
      hash: Binary.createFromBase64('gpGnPOB17p3NnKu74vhm+kCLXf8=', 0),
      keyId: Long('7446434162581438465')
    }
  },
  operationTime: Timestamp({ t: 1740781674, i: 1 })
}

db.embedded_movies.explain("allPlansExecution").aggregate([
  {
    "$vectorSearch": {
      "index": "vector_index",
      "path": "plot_embedding",
      "queryVector": QUERY_EMBEDDING,
      "exact": true,
      "limit": 10
    }
  }
])

{
  explainVersion: '1',
  stages: [
    {
      '$vectorSearch': {
        index: 'vector_index',
        path: 'plot_embedding',
        queryVector: [
            -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
            ...
            -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
        ],
        exact: true,
        limit: 10,
        explain: {
          query: {
            type: 'ExactVectorSearchQuery',
            args: {
              field: '$type:knnVector/plot_embedding',
              similarityFunction: 'dotProduct',
              filter: {
                type: 'DefaultQuery',
                args: { queryType: 'FieldExistsQuery' },
                stats: {
                  context: {
                    millisElapsed: 0.031999,
                    invocationCounts: {
                      createWeight: Long('1'),
                      createScorer: Long('2')
                    }
                  },
                  match: {
                    millisElapsed: 1.2138,
                    invocationCounts: { nextDoc: Long('3403') }
                  },
                  score: { millisElapsed: 0 }
                }
              }
            },
            stats: {
              context: {
                millisElapsed: 0.055318,
                invocationCounts: { createWeight: Long('1'), createScorer: Long('2') }
              },
              match: {
                millisElapsed: 2.578062,
                invocationCounts: { nextDoc: Long('3403') }
              },
              score: {
                millisElapsed: 5.624325,
                invocationCounts: {
                  score: Long('3402'),
                  setMinCompetitiveScore: Long('71')
                }
              }
            }
          },
          collectors: {
            allCollectorStats: {
              millisElapsed: 7.095773,
              invocationCounts: {
                collect: Long('3402'),
                competitiveIterator: Long('1'),
                setScorer: Long('1')
              }
            },
            facet: { collectorStats: { millisElapsed: 0 } }
          },
          metadata: {
            mongotVersion: '1.44.0',
            mongotHostName: '<hostname>.mongodb.net',
            indexName: 'vector_index',
            totalLuceneDocs: 3483
          },
          resourceUsage: {
            majorFaults: Long('0'),
            minorFaults: Long('0'),
            userTimeMs: Long('0'),
            systemTimeMs: Long('0'),
            maxReportingThreads: 1,
            numBatches: 1
          }
        }
      },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('19')
    },
    {
      '$_internalSearchIdLookup': { limit: Long('10') },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('19')
    }
  ],
  queryShapeHash: '9EFFBE6AC5B8DC66D20DB8939CF3ED6669D0C09BB6938C17AD3FB8C7793765B8',
  serverInfo: {
    host: '<hostname>.mongodb.net',
    port: 27017,
    version: '8.0.5',
    gitVersion: 'cb9e2e5e552ee39dea1e39d7859336456d0c9820'
  },
  serverParameters: {
    internalQueryFacetBufferSizeBytes: 104857600,
    internalQueryFacetMaxOutputDocSizeBytes: 104857600,
    internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
    internalDocumentSourceGroupMaxMemoryBytes: 104857600,
    internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
    internalQueryProhibitBlockingMergeOnMongoS: 0,
    internalQueryMaxAddToSetBytes: 104857600,
    internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
    internalQueryFrameworkControl: 'trySbeRestricted',
    internalQueryPlannerIgnoreIndexWithCollationForRegex: 1
  },
  command: {
    aggregate: 'embedded_movies',
    pipeline: [
      {
        '$vectorSearch': {
          index: 'vector_index',
          path: 'plot_embedding',
          queryVector: [
              -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
              ...
              -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
          ],
          exact: true,
          limit: 10
        }
      }
    ],
    cursor: {},
    '$db': 'sample_mflix'
  },
  ok: 1,
  '$clusterTime': {
    clusterTime: Timestamp({ t: 1740781894, i: 1 }),
    signature: {
      hash: Binary.createFromBase64('cNmipB7AsFGaT4ipmWCKizM/EbU=', 0),
      keyId: Long('7446434162581438465')
    }
  },
  operationTime: Timestamp({ t: 1740781894, i: 1 })
}

`queryPlanner`

The following example runs the explain method on an ANN and ENN vector search query for the plot_embedding field with the queryPlanner verbosity mode.

db.embedded_movies.explain("queryPlanner").aggregate([
  {
    "$vectorSearch": {
      "index": "vector_index",
      "path": "plot_embedding",
      "queryVector": QUERY_EMBEDDING,
      "numCandidates": 150,
      "limit": 10
    }
  }
])

{
  explainVersion: '1',
  stages: [
    {
      '$vectorSearch': {
        index: 'vector_index',
        path: 'plot_embedding',
        queryVector: [
            -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
            ...
            -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
        ],
        numCandidates: 150,
        limit: 10,
        explain: {
          query: {
            type: 'WrappedKnnQuery',
            args: {
              query: [
                {
                  type: 'KnnFloatVectorQuery',
                  args: { field: '$type:knnVector/plot_embedding', k: 150 }
                },
                { type: 'DocAndScoreQuery', args: {} }
              ]
            }
          },
          metadata: {
            mongotVersion: '1.44.0',
            mongotHostName: '<hostname>.mongodb.net',
            indexName: 'vector_index',
            totalLuceneDocs: 3483
          }
        }
      }
    },
    { '$_internalSearchIdLookup': { limit: Long('10') } }
  ],
  queryShapeHash: '9EFFBE6AC5B8DC66D20DB8939CF3ED6669D0C09BB6938C17AD3FB8C7793765B8',
  serverInfo: {
    host: '<hostname>.mongodb.net',
    port: 27017,
    version: '8.0.5',
    gitVersion: 'cb9e2e5e552ee39dea1e39d7859336456d0c9820'
  },
  serverParameters: {
    internalQueryFacetBufferSizeBytes: 104857600,
    internalQueryFacetMaxOutputDocSizeBytes: 104857600,
    internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
    internalDocumentSourceGroupMaxMemoryBytes: 104857600,
    internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
    internalQueryProhibitBlockingMergeOnMongoS: 0,
    internalQueryMaxAddToSetBytes: 104857600,
    internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
    internalQueryFrameworkControl: 'trySbeRestricted',
    internalQueryPlannerIgnoreIndexWithCollationForRegex: 1
  },
  command: {
    aggregate: 'embedded_movies',
    pipeline: [
      {
        '$vectorSearch': {
          index: 'vector_index',
          path: 'plot_embedding',
          queryVector: [
              -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
              ...
              -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
          ],
          numCandidates: 150,
          limit: 10
        }
      }
    ],
    cursor: {},
    '$db': 'sample_mflix'
  },
  ok: 1,
  '$clusterTime': {
    clusterTime: Timestamp({ t: 1740782044, i: 1 }),
    signature: {
      hash: Binary.createFromBase64('/ZHh6h+2qbJtYdOBHmeZepo1Wsc=', 0),
      keyId: Long('7446434162581438465')
    }
  },
  operationTime: Timestamp({ t: 1740782044, i: 1 })
}

db.embedded_movies.explain("queryPlanner").aggregate([
  {
    "$vectorSearch": {
      "index": "vector_index",
      "path": "plot_embedding",
      "queryVector": QUERY_EMBEDDING,
      "exact": true,
      "limit": 10
    }
  }
])

{
  explainVersion: '1',
  stages: [
    {
      '$vectorSearch': {
        index: 'vector_index',
        path: 'plot_embedding',
        queryVector: [
            -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
            ...
            -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
        ],
        exact: true,
        limit: 10,
        explain: {
          query: {
            type: 'ExactVectorSearchQuery',
            args: {
              field: '$type:knnVector/plot_embedding',
              similarityFunction: 'dotProduct',
              filter: {
                type: 'DefaultQuery',
                args: { queryType: 'FieldExistsQuery' }
              }
            }
          },
          metadata: {
            mongotVersion: '1.44.0',
            mongotHostName: '<hostname>.mongodb.net',
            indexName: 'vector_index',
            totalLuceneDocs: 3483
          }
        }
      }
    },
    { '$_internalSearchIdLookup': { limit: Long('10') } }
  ],
  queryShapeHash: '9EFFBE6AC5B8DC66D20DB8939CF3ED6669D0C09BB6938C17AD3FB8C7793765B8',
  serverInfo: {
    host: '<hostname>.mongodb.net',
    port: 27017,
    version: '8.0.5',
    gitVersion: 'cb9e2e5e552ee39dea1e39d7859336456d0c9820'
  },
  serverParameters: {
    internalQueryFacetBufferSizeBytes: 104857600,
    internalQueryFacetMaxOutputDocSizeBytes: 104857600,
    internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
    internalDocumentSourceGroupMaxMemoryBytes: 104857600,
    internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
    internalQueryProhibitBlockingMergeOnMongoS: 0,
    internalQueryMaxAddToSetBytes: 104857600,
    internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
    internalQueryFrameworkControl: 'trySbeRestricted',
    internalQueryPlannerIgnoreIndexWithCollationForRegex: 1
  },
  command: {
    aggregate: 'embedded_movies',
    pipeline: [
      {
        '$vectorSearch': {
          index: 'vector_index',
          path: 'plot_embedding',
          queryVector: [
              -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
              ...
              -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
          ],
          exact: true,
          limit: 10
        }
      }
    ],
    cursor: {},
    '$db': 'sample_mflix'
  },
  ok: 1,
  '$clusterTime': {
    clusterTime: Timestamp({ t: 1740782103, i: 23 }),
    signature: {
      hash: Binary.createFromBase64('yn3TtM4cfhu6HXCGzBGPbLmlM4E=', 0),
      keyId: Long('7446434162581438465')
    }
  },
  operationTime: Timestamp({ t: 1740782103, i: 23 })
}

`executionStats`

The following example runs the explain method on an ANN and ENN vector search query for the plot_embedding field with the executionStats verbosity mode.

db.embedded_movies.explain("executionStats").aggregate([
  {
    "$vectorSearch": {
      "index": "vector_index",
      "path": "plot_embedding",
      "queryVector": QUERY_EMBEDDING,
      "numCandidates": 150,
      "limit": 10
    }
  }
])

{
  explainVersion: '1',
  stages: [
    {
      '$vectorSearch': {
        index: 'vector_index',
        path: 'plot_embedding',
        queryVector: [
            -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
            ...
            -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
        ],
        numCandidates: 150,
        limit: 10,
        explain: {
          query: {
            type: 'WrappedKnnQuery',
            args: {
              query: [
                {
                  type: 'KnnFloatVectorQuery',
                  args: { field: '$type:knnVector/plot_embedding', k: 150 },
                  stats: {
                    context: { millisElapsed: 0 },
                    match: { millisElapsed: 0 },
                    score: { millisElapsed: 0 }
                  }
                },
                {
                  type: 'DocAndScoreQuery',
                  args: {},
                  stats: {
                    context: {
                      millisElapsed: 0.015853,
                      invocationCounts: {
                        createWeight: Long('1'),
                        createScorer: Long('2')
                      }
                    },
                    match: {
                      millisElapsed: 0.013112,
                      invocationCounts: { nextDoc: Long('151') }
                    },
                    score: {
                      millisElapsed: 0.033269,
                      invocationCounts: {
                        score: Long('150'),
                        setMinCompetitiveScore: Long('30')
                      }
                    }
                  }
                }
              ]
            },
            stats: {
              context: {
                millisElapsed: 3.979414,
                invocationCounts: {
                  vectorExecution: Long('1'),
                  createWeight: Long('1'),
                  createScorer: Long('2')
                }
              },
              match: {
                millisElapsed: 0.013112,
                invocationCounts: { nextDoc: Long('151') }
              },
              score: {
                millisElapsed: 0.033269,
                invocationCounts: {
                  score: Long('150'),
                  setMinCompetitiveScore: Long('30')
                }
              }
            }
          },
          collectors: {
            allCollectorStats: {
              millisElapsed: 0.115101,
              invocationCounts: {
                collect: Long('150'),
                competitiveIterator: Long('1'),
                setScorer: Long('1')
              }
            },
            facet: { collectorStats: { millisElapsed: 0 } }
          },
          metadata: {
            mongotVersion: '1.44.0',
            mongotHostName: '<hostname>.mongodb.net',
            indexName: 'vector_index',
            totalLuceneDocs: 3483
          },
          resourceUsage: {
            majorFaults: Long('0'),
            minorFaults: Long('0'),
            userTimeMs: Long('0'),
            systemTimeMs: Long('0'),
            maxReportingThreads: 1,
            numBatches: 1
          }
        }
      },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('13')
    },
    {
      '$_internalSearchIdLookup': { limit: Long('10') },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('13')
    }
  ],
  queryShapeHash: '9EFFBE6AC5B8DC66D20DB8939CF3ED6669D0C09BB6938C17AD3FB8C7793765B8',
  serverInfo: {
    host: '<hostname>.mongodb.net',
    port: 27017,
    version: '8.0.5',
    gitVersion: 'cb9e2e5e552ee39dea1e39d7859336456d0c9820'
  },
  serverParameters: {
    internalQueryFacetBufferSizeBytes: 104857600,
    internalQueryFacetMaxOutputDocSizeBytes: 104857600,
    internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
    internalDocumentSourceGroupMaxMemoryBytes: 104857600,
    internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
    internalQueryProhibitBlockingMergeOnMongoS: 0,
    internalQueryMaxAddToSetBytes: 104857600,
    internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
    internalQueryFrameworkControl: 'trySbeRestricted',
    internalQueryPlannerIgnoreIndexWithCollationForRegex: 1
  },
  command: {
    aggregate: 'embedded_movies',
    pipeline: [
      {
        '$vectorSearch': {
          index: 'vector_index',
          path: 'plot_embedding',
          queryVector: [
              -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
              ...
              -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
          ],
          numCandidates: 150,
          limit: 10
        }
      }
    ],
    cursor: {},
    '$db': 'sample_mflix'
  },
  ok: 1,
  '$clusterTime': {
    clusterTime: Timestamp({ t: 1740782136, i: 1 }),
    signature: {
      hash: Binary.createFromBase64('33LuqO946oIzmslClhuu+Z+PthQ=', 0),
      keyId: Long('7446434162581438465')
    }
  },
  operationTime: Timestamp({ t: 1740782136, i: 1 })
}

db.embedded_movies.explain("executionStats").aggregate([
  {
    "$vectorSearch": {
      "index": "vector_index",
      "path": "plot_embedding",
      "queryVector": QUERY_EMBEDDING,
      "exact": true,
      "limit": 10
    }
  }
])

{
  explainVersion: '1',
  stages: [
    {
      '$vectorSearch': {
        index: 'vector_index',
        path: 'plot_embedding',
        queryVector: [
            -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
            ...
            -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
        ],
        exact: true,
        limit: 10,
        explain: {
          query: {
            type: 'ExactVectorSearchQuery',
            args: {
              field: '$type:knnVector/plot_embedding',
              similarityFunction: 'dotProduct',
              filter: {
                type: 'DefaultQuery',
                args: { queryType: 'FieldExistsQuery' },
                stats: {
                  context: {
                    millisElapsed: 0.029472,
                    invocationCounts: {
                      createWeight: Long('1'),
                      createScorer: Long('2')
                    }
                  },
                  match: {
                    millisElapsed: 0.234054,
                    invocationCounts: { nextDoc: Long('3403') }
                  },
                  score: { millisElapsed: 0 }
                }
              }
            },
            stats: {
              context: {
                millisElapsed: 0.050328,
                invocationCounts: { createWeight: Long('1'), createScorer: Long('2') }
              },
              match: {
                millisElapsed: 0.622028,
                invocationCounts: { nextDoc: Long('3403') }
              },
              score: {
                millisElapsed: 3.03997,
                invocationCounts: {
                  score: Long('3402'),
                  setMinCompetitiveScore: Long('71')
                }
              }
            }
          },
          collectors: {
            allCollectorStats: {
              millisElapsed: 3.601357,
              invocationCounts: {
                collect: Long('3402'),
                competitiveIterator: Long('1'),
                setScorer: Long('1')
              }
            },
            facet: { collectorStats: { millisElapsed: 0 } }
          },
          metadata: {
            mongotVersion: '1.44.0',
            mongotHostName: '<hostname>.mongodb.net',
            indexName: 'vector_index',
            totalLuceneDocs: 3483
          },
          resourceUsage: {
            majorFaults: Long('0'),
            minorFaults: Long('0'),
            userTimeMs: Long('0'),
            systemTimeMs: Long('0'),
            maxReportingThreads: 1,
            numBatches: 1
          }
        }
      },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('12')
    },
    {
      '$_internalSearchIdLookup': { limit: Long('10') },
      nReturned: Long('0'),
      executionTimeMillisEstimate: Long('12')
    }
  ],
  queryShapeHash: '9EFFBE6AC5B8DC66D20DB8939CF3ED6669D0C09BB6938C17AD3FB8C7793765B8',
  serverInfo: {
    host: '<hostname>.mongodb.net',
    port: 27017,
    version: '8.0.5',
    gitVersion: 'cb9e2e5e552ee39dea1e39d7859336456d0c9820'
  },
  serverParameters: {
    internalQueryFacetBufferSizeBytes: 104857600,
    internalQueryFacetMaxOutputDocSizeBytes: 104857600,
    internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
    internalDocumentSourceGroupMaxMemoryBytes: 104857600,
    internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
    internalQueryProhibitBlockingMergeOnMongoS: 0,
    internalQueryMaxAddToSetBytes: 104857600,
    internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
    internalQueryFrameworkControl: 'trySbeRestricted',
    internalQueryPlannerIgnoreIndexWithCollationForRegex: 1
  },
  command: {
    aggregate: 'embedded_movies',
    pipeline: [
      {
        '$vectorSearch': {
          index: 'vector_index',
          path: 'plot_embedding',
          queryVector: [
              -0.0016261312,   -0.028070757,   -0.011342932,    -0.012775794,
              ...
              -0.0060913274,   -0.007130985,   -0.013109017,    -0.009710136
          ],
          exact: true,
          limit: 10
        }
      }
    ],
    cursor: {},
    '$db': 'sample_mflix'
  },
  ok: 1,
  '$clusterTime': {
    clusterTime: Timestamp({ t: 1740782184, i: 1 }),
    signature: {
      hash: Binary.createFromBase64('tpxoW70Y+Z5HmDqaswag+matLQQ=', 0),
      keyId: Long('7446434162581438465')
    }
  },
  operationTime: Timestamp({ t: 1740782184, i: 1 })
}

Back

Create and Run Queries

Transform Documents & Filter Collections