8.26 V$KEY_VECTOR

V$KEY_VECTOR provides debugging information related to the data structures used by in-memory aggregation for current and recent queries using key vectors.

Column Datatype Description

SESSION_ID

NUMBER

Session ID

CON_ID

NUMBER

The ID of the container to which the data pertains. Possible values include:

  • 0: This value is used for rows containing data that pertain to the entire CDB. This value is also used for rows in non-CDBs.

  • 1: This value is used for rows containing data that pertain to only the root

  • n: Where n is the applicable container ID for the rows containing data

TRANSLATE_ID

NUMBER

Translation vector ID

SQL_ID

VARCHAR2(13)

SQL ID that uses the translation vector

SQL_EXEC_START

DATE

Time when the execution of the SQL started

SQL_EXEC_ID

NUMBER

SQL execution identifier

PROCESS

VARCHAR2(64)

Operating system client process ID

STATE

VARCHAR2(8)

State of the in-memory aggregation operation:

  • BYPASS

    Refers to key vectors that are currently in use by queries that have undergone the vector transform.

  • BYPASSED

    This is a historical entry for key vectors whose queries have already finished.

  • FINISHED

    This is a historical entry for key vectors whose queries have already finished.

  • RUNNING

    Refers to key vectors that are currently in use by queries that have undergone the vector transform.

One query will probably have multiple key vectors if it has gone through the vector transform.

Also, if the query is operating in parallel, there may be multiple entries per PQ slave.

The PROCESS column can be joined to V$SESSION to differentiate these cases.

TYPE

VARCHAR2(8)

Type of translation vector created:

  • DOUBLEIND

  • PAGED

  • SIMPLE

  • INDIRECT

  • HASH

  • OFFSET

WIDTH

NUMBER

Width (number of bits) of DOUBLEIND, SIMPLE, INDIRECT, or OFFSETvector translation array:

  • 1

  • 4

  • 8

  • 16

  • 32

KEY_DATA_TYPE

VARCHAR2(13)

The internal join column key data type of either DOUBLEIND, SIMPLE, INDIRECT, or OFFSET translation vectors:

  • BINARY

  • BINARY_FLOAT

  • NUMBER

  • PACKED_BINARY

  • PACKED_DATE

  • PACKED_NUMBER

Also, a value of NULL is displayed for HASH translation vectors with non-numeric data types and for HASH translation vectors with more than one join key column.

JOIN_COLUMN_COUNT

NUMBER

Number of join columns. PAGED and HASHED are the only types which can have greater than one join column.

JOIN_KEY_COUNT

NUMBER

Number of join key values used to build translation vector

DUP_JOIN_KEY_COUNT

NUMBER

Indicates the number of key values in the key vector that have more than one parent value

MIN_JOIN_KEY

NUMBER

Min join key value in translation vector

MAX_JOIN_KEY

NUMBER

Max join key value in translation vector

GROUP_KEY_COUNT

NUMBER

Max dense grouping key value. This value is computed while the data that creates the translation vector is processed.

FILTERED

NUMBER

Number of rows filtered by translation vector

PROBED

NUMBER

Number of rows that probed the translation vector in key vector use row source

ACTIVE

NUMBER

Number of active translation vectors used as filters across all slaves

DISABLED

NUMBER

Number of translation vector filters across all slaves that were disabled

MEMORY_ALLOCATED

NUMBER

Amount of memory allocated for the key vector

JOIN_STRUCTURE_SIZE

NUMBER

Amount of memory used out of the allocated space

FACT_OWNER

VARCHAR2(128)

The owner of the fact table. Null when the fact table is null.

FACT_NAME

VARCHAR2(128)

The table that contains measure data. May be null if more than one fact table is used in the query.

DIMENSION_OWNER

VARCHAR2(128)

When a single table is used to construct the key vector, the table owner will appear in this column. If multiple tables are joined to serve as the dimension (more of a snowflake than a star schema shape, for one example), then this column will be null.

DIMENSION_NAME

VARCHAR2(128)

The table that contains attribute data and is joined to the fact table. May be null if two or more tables are used (for example, snowflake style dimension tables).

CREATION_DURATION

NUMBER

Records the total time in seconds that it took to create the key vector

PAYLOAD_COLUMN_COUNT

NUMBER

The number of columns being carried from the dimension table to the fact table scan for processing without joinback

See Also:

Oracle Database SQL Tuning Guide for more information about in-memory aggregation