[Hive - LanguageManual ] Explain (待)

EXPLAIN Syntax

EXPLAIN Syntax

Hive provides an EXPLAIN command that shows the execution plan for a query. The syntax for this statement is as follows:

EXPLAIN [EXTENDED|DEPENDENCY|AUTHORIZATION] query

AUTHORIZATION is supported from HIVE 0.14.0 via HIVE-5961.

The use of EXTENDED in the EXPLAIN statement produces extra information about the operators in the plan. This is typically physical information like file names.

A Hive query gets converted into a sequence (it is more an Directed Acyclic Graph) of stages. These stages may be map/reduce stages or they may even be stages that do metastore or file system operations like move and rename. The explain output comprises of three parts:

The Abstract Syntax Tree for the query
The dependencies between the different stages of the plan
The description of each of the stages

The description of the stages itself shows a sequence of operators with the metadata associated with the operators. The metadata may comprise of things like filter expressions for the FilterOperator or the select expressions for the SelectOperator or the output file names for the FileSinkOperator.

As an example, consider the following EXPLAIN query:

EXPLAIN
FROM src INSERT OVERWRITE TABLE dest_g1 SELECT src.key, sum(substr(src.value,4)) GROUP BY src.key;

The output of this statement contains the following parts:

The Abstract Syntax Tree

ABSTRACT SYNTAX TREE:
  (TOK_QUERY (TOK_FROM (TOK_TABREF src)) (TOK_INSERT (TOK_DESTINATION (TOK_TAB dest_g1)) (TOK_SELECT (TOK_SELEXPR (TOK_COLREF src key)) (TOK_SELEXPR (TOK_FUNCTION sum (TOK_FUNCTION substr (TOK_COLREF src value) 4)))) (TOK_GROUPBY (TOK_COLREF src key))))

The Dependency Graph

STAGE DEPENDENCIES:

  Stage-1 is a root stage

  Stage-2 depends on stages: Stage-1

  Stage-0 depends on stages: Stage-2

This shows that Stage-1 is the root stage, Stage-2 is executed after Stage-1 is done and Stage-0 is executed after Stage-2 is done.

The plans of each Stage

STAGE PLANS:
  Stage: Stage-1
    Map Reduce
      Alias -> Map Operator Tree:
        src
            Reduce Output Operator
              key expressions:
                    expr: key
                    type: string
              sort order: +
              Map-reduce partition columns:
                    expr: rand()
                    type: double
              tag: -1
              value expressions:
                    expr: substr(value, 4)
                    type: string
      Reduce Operator Tree:
        Group By Operator
          aggregations:
                expr: sum(UDFToDouble(VALUE.0))
          keys:
                expr: KEY.0
                type: string
          mode: partial1
          File Output Operator
            compressed: false
            table:
                input format: org.apache.hadoop.mapred.SequenceFileInputFormat
                output format: org.apache.hadoop.mapred.SequenceFileOutputFormat
                name: binary_table
 
  Stage: Stage-2
    Map Reduce
      Alias -> Map Operator Tree:
        /tmp/hive-zshao/67494501/106593589.10001
          Reduce Output Operator
            key expressions:
                  expr: 0
                  type: string
            sort order: +
            Map-reduce partition columns:
                  expr: 0
                  type: string
            tag: -1
            value expressions:
                  expr: 1
                  type: double
      Reduce Operator Tree:
        Group By Operator
          aggregations:
                expr: sum(VALUE.0)
          keys:
                expr: KEY.0
                type: string
          mode: final
          Select Operator
            expressions:
                  expr: 0
                  type: string
                  expr: 1
                  type: double
            Select Operator
              expressions:
                    expr: UDFToInteger(0)
                    type: int
                    expr: 1
                    type: double
              File Output Operator
                compressed: false
                table:
                    input format: org.apache.hadoop.mapred.TextInputFormat
                    output format: org.apache.hadoop.hive.ql.io.IgnoreKeyTextOutputFormat
                    serde: org.apache.hadoop.hive.serde2.dynamic_type.DynamicSerDe
                    name: dest_g1
 
  Stage: Stage-0
    Move Operator
      tables:
            replace: true
            table:
                input format: org.apache.hadoop.mapred.TextInputFormat
                output format: org.apache.hadoop.hive.ql.io.IgnoreKeyTextOutputFormat
                serde: org.apache.hadoop.hive.serde2.dynamic_type.DynamicSerDe
                name: dest_g1

In this example there are 2 map/reduce stages (Stage-1 and Stage-2) and 1 File System related stage (Stage-0). Stage-0 basically moves the results from a temporary directory to the directory corresponding to the table dest_g1.

A map/reduce stage itself comprises of 2 parts:

A mapping from table alias to Map Operator Tree - This mapping tells the mappers which operator tree to call in order to process the rows from a particular table or result of a previous map/reduce stage. In Stage-1 in the above example, the rows from src table are processed by the operator tree rooted at a Reduce Output Operator. Similarly, in Stage-2 the rows of the results of Stage-1 are processed by another operator tree rooted at another Reduce Output Operator. Each of these Reduce Output Operators partitions the data to the reducers according to the criteria shown in the metadata.
A Reduce Operator Tree - This is the operator tree which processes all the rows on the reducer of the map/reduce job. In Stage-1 for example, the Reducer Operator Tree is carrying out a partial aggregation where as the Reducer Operator Tree in Stage-2 computes the final aggregation from the partial aggregates computed in Stage-1

The use of DEPENDENCY in the EXPLAIN statement produces extra information about the inputs in the plan. It shows various attributes for the inputs. For example, for a query like:

EXPLAIN DEPENDENCY
  SELECT key, count(1) FROM srcpart WHERE ds IS NOT NULL GROUP BY key

the following output is produced:

{"input_partitions":[{"partitionName":"default<at:var at:name="srcpart" />ds=2008-04-08/hr=11"},{"partitionName":"default<at:var at:name="srcpart" />ds=2008-04-08/hr=12"},{"partitionName":"default<at:var at:name="srcpart" />ds=2008-04-09/hr=11"},{"partitionName":"default<at:var at:name="srcpart" />ds=2008-04-09/hr=12"}],"input_tables":[{"tablename":"default@srcpart","tabletype":"MANAGED_TABLE"}]}

The inputs contain both the tables and the partitions. Note that the table is present even if none of the partitions is accessed in the query.

The dependencies show the parents in case a table is accessed via a view. Consider the following queries:

CREATE VIEW V1 AS SELECT key, value from src;
EXPLAIN DEPENDENCY SELECT * FROM V1;

The following output is produced:

{"input_partitions":[],"input_tables":[{"tablename":"default@v1","tabletype":"VIRTUAL_VIEW"},{"tablename":"default@src","tabletype":"MANAGED_TABLE","tableParents":"[default@v1]"}]}

As above, the inputs contain the view V1 and the table ‘src‘ that the view V1 refers to.

All the outputs are shown if a table is being accessed via multiple parents.

CREATE VIEW V2 AS SELECT ds, key, value FROM srcpart WHERE ds IS NOT NULL;
CREATE VIEW V4 AS
  SELECT src1.key, src2.value as value1, src3.value as value2
  FROM V1 src1 JOIN V2 src2 on src1.key = src2.key JOIN src src3 ON src2.key = src3.key;
EXPLAIN DEPENDENCY SELECT * FROM V4;

The following output is produced.

{"input_partitions":[{"partitionParents":"[default@v2]","partitionName":"default<at:var at:name="srcpart" />ds=2008-04-08/hr=11"},{"partitionParents":"[default@v2]","partitionName":"default<at:var at:name="srcpart" />ds=2008-04-08/hr=12"},{"partitionParents":"[default@v2]","partitionName":"default<at:var at:name="srcpart" />ds=2008-04-09/hr=11"},{"partitionParents":"[default@v2]","partitionName":"default<at:var at:name="srcpart" />ds=2008-04-09/hr=12"}],"input_tables":[{"tablename":"default@v4","tabletype":"VIRTUAL_VIEW"},{"tablename":"default@v2","tabletype":"VIRTUAL_VIEW","tableParents":"[default@v4]"},{"tablename":"default@v1","tabletype":"VIRTUAL_VIEW","tableParents":"[default@v4]"},{"tablename":"default@src","tabletype":"MANAGED_TABLE","tableParents":"[default@v4, default@v1]"},{"tablename":"default@srcpart","tabletype":"MANAGED_TABLE","tableParents":"[default@v2]"}]}

As can be seen, src is being accessed via parents v1 and v4.

The use of AUTHORIZATION in the EXPLAIN statement shows all entities needed to be authorized to execute the query and authorization failures if exists. For example, for a query like:

EXPLAIN AUTHORIZATION
  SELECT * FROM src JOIN srcpart;

the following output is produced:

INPUTS:
  default@srcpart
  default@src
  default@srcpart@ds=2008-04-08/hr=11
  default@srcpart@ds=2008-04-08/hr=12
  default@srcpart@ds=2008-04-09/hr=11
  default@srcpart@ds=2008-04-09/hr=12
OUTPUTS:
  hdfs://localhost:9000/tmp/.../-mr-10000
CURRENT_USER:
  navis
OPERATION:
  QUERY
AUTHORIZATION_FAILURES:
  Permission denied: Principal [name=navis, type=USER] does not have following privileges for operation QUERY [[SELECT] on Object [type=TABLE_OR_VIEW, name=default.src], [SELECT] on Object [type=TABLE_OR_VIEW, name=default.srcpart]]

With the FORMATTED keyword, it will be returned in JSON format.

"OUTPUTS":["hdfs://localhost:9000/tmp/.../-mr-10000"],"INPUTS":["default@srcpart","default@src","default@srcpart@ds=2008-04-08/hr=11","default@srcpart@ds=2008-04-08/hr=12","default@srcpart@ds=2008-04-09/hr=11","default@srcpart@ds=2008-04-09/hr=12"],"OPERATION":"QUERY","CURRENT_USER":"navis","AUTHORIZATION_FAILURES":["Permission denied: Principal [name=navis, type=USER] does not have following privileges for operation QUERY [[SELECT] on Object [type=TABLE_OR_VIEW, name=default.src], [SELECT] on Object [type=TABLE_OR_VIEW, name=default.srcpart]]"]}