Returns the query plan for a SQL Data Warehouse SQL Data Warehouse SQL statement without running the statement. Use EXPLAIN to preview which operations will require data movement and to view the estimated costs of the query operations.
For more information about query plans, see “Understanding Query Plans” in the [Parallel Data Warehouse product documentation](https://www.microsoft.com/en-us/download/details.aspx?id=51610)
EXPLAIN SQL_statement
[;] SQL_statement
The SQL statement on which EXPLAIN will run. SQL_statement can be any of these commands: SELECT, INSERT, UPDATE, DELETE, CREATE TABLE AS SELECT, CREATE REMOTE TABLE.
Requires the SHOWPLAN permission, and permission to execute SQL_statement. See Permissions: GRANT, DENY, REVOKE (Azure SQL Data Warehouse, Parallel Data Warehouse).
The return value from the EXPLAIN command is an XML document with the structure shown below. This XML document lists all operations in the query plan for the given query, each enclosed by the <dsql_operation> tag. The return value is of type nvarchar(max).
The returned query plan depicts sequential SQL statements; when the query runs it may involve parallelized operations, so some of the sequential statements shown may run at the same time.
\<?xml version="1.0" encoding="utf-8"?>
<dsql_query>
<sql>. . .</sql>
<params />
<dsql_operations>
<dsql_operation>
. . .
</dsql_operation>
[ . . . n ]
<dsql_operations>
</dsql_query> The XML tags contain this information:
| XML Tag | Summary, Attributes, and Content |
|---|---|
| <dsql_query> | Top level/document element. |
| <sql> | Echoes SQL_statement. |
| <params> | This tag is not used at this time. |
| <dsql_operations> | Summarizes and contains the query steps, and includes cost information for the query. Also contains all of the <dsql_operation> blocks. This tag contains count information for the entire query:<dsql_operations total_cost=total_cost total_number_operations=total_number_operations>total_cost is the total estimated time for the query to run, in ms. total_number_operations is the total number of operations for the query. An operation that will be parallelized and run on multiple nodes is counted as a single operation. |
| <dsql_operation> | Describes a single operation within the query plan. The <dsql_operation> tag contains the operation type as an attribute:<dsql_operation operation_type=operation_type>operation_type is one of the values found in Querying Data (SQL Server PDW). The content in the \<dsql_operation> block is dependent on the operation type.See the table below. |
| Operation Type | Content | Example |
|---|---|---|
| BROADCAST_MOVE, DISTRIBUTE_REPLICATED_TABLE_MOVE, MASTER_TABLE_MOVE, PARTITION_MOVE, SHUFFLE_MOVE, and TRIM_MOVE | <operation_cost> element, with these attributes. Values reflect only the local operation:- cost is the local operator cost and shows the estimated time for the operation to run, in ms. - accumulative_cost is the sum of all seen operations in the plan including summed values for parallel operations, in ms. - average_rowsize is the estimated average row size (in bytes) of rows retrieved and passed during the operation. - output_rows is the output (node) cardinality and shows the number of output rows. <location>: The nodes or distributions where the operation will occur. Options are: “Control”, “ComputeNode”, “AllComputeNodes”, “AllDistributions”, “SubsetDistributions”, “Distribution”, and “SubsetNodes”.<source_statement>: The source data for the shuffle move.<destination_table>: The internal temporary table the data will be moved into.<shuffle_columns>: (Applicable only to SHUFFLE_MOVE operations). One or more columns that will be used as the distribution columns for the temporary table. |
<operation_cost cost="40" accumulative_cost="40" average_rowsize = "50" output_rows="100"/><location distribution="AllDistributions" /><source_statement type="statement">SELECT [TableAlias_3b77ee1d8ccf4a94ba644118b355db9d].[dist_date] FROM [qatest].[dbo].[flyers] [TableAlias_3b77ee1d8ccf4a94ba644118b355db9d] </source_statement><destination_table>Q_[TEMP_ID_259]_[PARTITION_ID]</destination_table><shuffle_columns>dist_date;</shuffle_columns> |
| CopyOperation | <operation_cost>: See <operation_cost> above.<DestinationCatalog>: The destination node or nodes.<DestinationSchema>: The destination schema in DestinationCatalog.<DestinationTableName>: Name of the destination table or “TableName”.<DestinationDatasource>: The name or connection information for the destination datasource.<Username> and <Password>: These fields indicate that a username and password for the destination may be required.<BatchSize>: The batch size for the copy operation.<SelectStatement>: The select statement used to perform the copy.<distribution>: The distribution where the copy is performed. |
<operation_cost cost="0" accumulative_cost="0" average_rowsize="4" output_rows="1" /><DestinationCatalog>master</DestinationCatalog><DestinationSchema>dbo</DestinationSchema><DestinationTableName>[TableName]</DestinationTableName><DestinationDatasource>localhost, 8080</DestinationDatasource><Username>...</Username><Password>...</Password><BatchSize>6000</BatchSize><SelectStatement>SELECT T1_1.c1 AS c1 FROM [qatest].[dbo].[gigs] AS T1_1</SelectStatement><distribution>ControlNode</distribution> |
| MetaDataCreate_Operation | <source_table>: The source table for the operation.<destionation_table>: The destination table for the operation. |
<source_table>databases</source_table><destination_table>MetaDataCreateLandingTempTable</destination_table> |
| ON | <location>: See <location> above.<sql_operation>: Identifies the SQL command that will be performed on a node. |
<location permanent="false" distribution="AllDistributions">Compute</location><sql_operation type="statement">CREATE TABLE [tempdb].[dbo]. [Q_[TEMP_ID_259]]_ [PARTITION_ID]]]([dist_date] DATE) WITH (DISTRIBUTION = HASH([dist_date]),) </sql_operation> |
| RemoteOnOperation | <DestinationCatalog>: The destination catalog.<DestinationSchema>: The destination schema in DestinationCatalog.<DestinationTableName>: Name of the destination table or “TableName”.<DestinationDatasource>: Name of the destination datasource.<Username> and <Password>: These fields indicate that a username and password for the destination may be required.<CreateStatement>: The table creation statement for the destination database. |
<DestinationCatalog>master</DestinationCatalog><DestinationSchema>dbo</DestinationSchema><DestinationTableName>TableName</DestinationTableName><DestinationDatasource>DestDataSource</DestinationDatasource><Username>...</Username><Password>...</Password><CreateStatement>CREATE TABLE [master].[dbo].[TableName] ([col1] BIGINT) ON [PRIMARY] WITH(DATA_COMPRESSION=PAGE);</CreateStatement> |
| RETURN | <resultset>: The identifier for the result set. |
<resultset>RS_19</resultset> |
| RND_ID | <identifier>: The identifier for the object created. |
<identifier>TEMP_ID_260</identifier> |
EXPLAIN can be applied to optimizable queries only, which are queries that can be improved or modified based on the results of an EXPLAIN command. The supported EXPLAIN commands are listed above. Attempting to use EXPLAIN with an unsupported query type will either return an error or echo the query.
EXPLAIN is not supported in a user transaction.
The following example shows an EXPLAIN command run on a SELECT statement, and the XML result.
Submitting an EXPLAIN statement
The submitted command for this example is:
-- Uses AdventureWorks
EXPLAIN
SELECT CAST (AVG(YearlyIncome) AS int) AS AverageIncome,
CAST(AVG(FIS.SalesAmount) AS int) AS AverageSales,
G.StateProvinceName, T.SalesTerritoryGroup
FROM dbo.DimGeography AS G
JOIN dbo.DimSalesTerritory AS T
ON G.SalesTerritoryKey = T.SalesTerritoryKey
JOIN dbo.DimCustomer AS C
ON G.GeographyKey = C.GeographyKey
JOIN dbo.FactInternetSales AS FIS
ON C.CustomerKey = FIS.CustomerKey
WHERE T.SalesTerritoryGroup IN ('North America', 'Pacific')
AND Gender = 'F'
GROUP BY G.StateProvinceName, T.SalesTerritoryGroup
ORDER BY AVG(YearlyIncome) DESC;
GO After executing the statement using the EXPLAIN option, the message tab presents a single line titled explain, and starting with the XML text \<?xml version="1.0" encoding="utf-8"?> Click on the XML to open the entire text in an XML window. To better understand the following comments, you should turn on the display of line numbers in SSDT.
With the output appearing in the explain tab SSDT, on the TOOLS menu, select Options.
Expand the Text Editor section, expand XML, and then click General.
In the Display area, check Line numbers.
Click OK.
Example EXPLAIN output
The XML result of the EXPLAIN command with row numbers turned on is:
1 \<?xml version="1.0" encoding="utf-8"?>
2 <dsql_query>
3 <sql>SELECT CAST (AVG(YearlyIncome) AS int) AS AverageIncome,
4 CAST(AVG(FIS.SalesAmount) AS int) AS AverageSales,
5 G.StateProvinceName, T.SalesTerritoryGroup
6 FROM dbo.DimGeography AS G
7 JOIN dbo.DimSalesTerritory AS T
8 ON G.SalesTerritoryKey = T.SalesTerritoryKey
9 JOIN dbo.DimCustomer AS C
10 ON G.GeographyKey = C.GeographyKey
11 JOIN dbo.FactInternetSales AS FIS
12 ON C.CustomerKey = FIS.CustomerKey
13 WHERE T.SalesTerritoryGroup IN ('North America', 'Pacific')
14 AND Gender = 'F'
15 GROUP BY G.StateProvinceName, T.SalesTerritoryGroup
16 ORDER BY AVG(YearlyIncome) DESC</sql>
17 <dsql_operations total_cost="0.926237696" total_number_operations="9">
18 <dsql_operation operation_type="RND_ID">
19 <identifier>TEMP_ID_16893</identifier>
20 </dsql_operation>
21 <dsql_operation operation_type="ON">
22 <location permanent="false" distribution="AllComputeNodes" />
23 <sql_operations>
24 <sql_operation type="statement">CREATE TABLE [tempdb].[dbo].[TEMP_ID_16893] ([CustomerKey] INT NOT NULL, [GeographyKey] INT, [YearlyIncome] MONEY ) WITH(DATA_COMPRESSION=PAGE);</sql_operation>
25 </sql_operations>
26 </dsql_operation>
27 <dsql_operation operation_type="BROADCAST_MOVE">
28 <operation_cost cost="0.121431552" accumulative_cost="0.121431552" average_rowsize="16" output_rows="31.6228" />
29 <source_statement>SELECT [T1_1].[CustomerKey] AS [CustomerKey],
30 [T1_1].[GeographyKey] AS [GeographyKey],
31 [T1_1].[YearlyIncome] AS [YearlyIncome]
32 FROM (SELECT [T2_1].[CustomerKey] AS [CustomerKey],
33 [T2_1].[GeographyKey] AS [GeographyKey],
34 [T2_1].[YearlyIncome] AS [YearlyIncome]
35 FROM [AdventureWorksPDW2012].[dbo].[DimCustomer] AS T2_1
36 WHERE ([T2_1].[Gender] = CAST (N'F' COLLATE Latin1_General_100_CI_AS_KS_WS AS NVARCHAR (1)) COLLATE Latin1_General_100_CI_AS_KS_WS)) AS T1_1</source_statement>
37 <destination_table>[TEMP_ID_16893]</destination_table>
38 </dsql_operation>
39 <dsql_operation operation_type="RND_ID">
40 <identifier>TEMP_ID_16894</identifier>
41 </dsql_operation>
42 <dsql_operation operation_type="ON">
43 <location permanent="false" distribution="AllDistributions" />
44 <sql_operations>
45 <sql_operation type="statement">CREATE TABLE [tempdb].[dbo].[TEMP_ID_16894] ([StateProvinceName] NVARCHAR(50) COLLATE Latin1_General_100_CI_AS_KS_WS, [SalesTerritoryGroup] NVARCHAR(50) COLLATE Latin1_General_100_CI_AS_KS_WS NOT NULL, [col] BIGINT, [col1] MONEY NOT NULL, [col2] BIGINT, [col3] MONEY NOT NULL ) WITH(DATA_COMPRESSION=PAGE);</sql_operation>
46 </sql_operations>
47 </dsql_operation>
48 <dsql_operation operation_type="SHUFFLE_MOVE">
49 <operation_cost cost="0.804806144" accumulative_cost="0.926237696" average_rowsize="232" output_rows="108.406" />
50 <source_statement>SELECT [T1_1].[StateProvinceName] AS [StateProvinceName],
51 [T1_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup],
52 [T1_1].[col2] AS [col],
53 [T1_1].[col] AS [col1],
54 [T1_1].[col3] AS [col2],
55 [T1_1].[col1] AS [col3]
56 FROM (SELECT ISNULL([T2_1].[col1], CONVERT (MONEY, 0.00, 0)) AS [col],
57 ISNULL([T2_1].[col3], CONVERT (MONEY, 0.00, 0)) AS [col1],
58 [T2_1].[StateProvinceName] AS [StateProvinceName],
59 [T2_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup],
60 [T2_1].[col] AS [col2],
61 [T2_1].[col2] AS [col3]
62 FROM (SELECT COUNT_BIG([T3_2].[YearlyIncome]) AS [col],
63 SUM([T3_2].[YearlyIncome]) AS [col1],
64 COUNT_BIG(CAST ((0) AS INT)) AS [col2],
65 SUM([T3_2].[SalesAmount]) AS [col3],
66 [T3_2].[StateProvinceName] AS [StateProvinceName],
67 [T3_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup]
68 FROM (SELECT [T4_1].[SalesTerritoryKey] AS [SalesTerritoryKey],
69 [T4_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup]
70 FROM [AdventureWorksPDW2012].[dbo].[DimSalesTerritory] AS T4_1
71 WHERE (([T4_1].[SalesTerritoryGroup] = CAST (N'North America' COLLATE Latin1_General_100_CI_AS_KS_WS AS NVARCHAR (13)) COLLATE Latin1_General_100_CI_AS_KS_WS)
72 OR ([T4_1].[SalesTerritoryGroup] = CAST (N'Pacific' COLLATE Latin1_General_100_CI_AS_KS_WS AS NVARCHAR (7)) COLLATE Latin1_General_100_CI_AS_KS_WS))) AS T3_1
73 INNER JOIN
74 (SELECT [T4_1].[SalesTerritoryKey] AS [SalesTerritoryKey],
75 [T4_2].[YearlyIncome] AS [YearlyIncome],
76 [T4_2].[SalesAmount] AS [SalesAmount],
77 [T4_1].[StateProvinceName] AS [StateProvinceName]
78 FROM [AdventureWorksPDW2012].[dbo].[DimGeography] AS T4_1
79 INNER JOIN
80 (SELECT [T5_2].[GeographyKey] AS [GeographyKey],
81 [T5_2].[YearlyIncome] AS [YearlyIncome],
82 [T5_1].[SalesAmount] AS [SalesAmount]
83 FROM [AdventureWorksPDW2012].[dbo].[FactInternetSales] AS T5_1
84 INNER JOIN
85 [tempdb].[dbo].[TEMP_ID_16893] AS T5_2
86 ON ([T5_1].[CustomerKey] = [T5_2].[CustomerKey])) AS T4_2
87 ON ([T4_2].[GeographyKey] = [T4_1].[GeographyKey])) AS T3_2
88 ON ([T3_1].[SalesTerritoryKey] = [T3_2].[SalesTerritoryKey])
89 GROUP BY [T3_2].[StateProvinceName], [T3_1].[SalesTerritoryGroup]) AS T2_1) AS T1_1</source_statement>
90 <destination_table>[TEMP_ID_16894]</destination_table>
91 <shuffle_columns>StateProvinceName;</shuffle_columns>
92 </dsql_operation>
93 <dsql_operation operation_type="ON">
94 <location permanent="false" distribution="AllComputeNodes" />
95 <sql_operations>
96 <sql_operation type="statement">DROP TABLE [tempdb].[dbo].[TEMP_ID_16893]</sql_operation>
97 </sql_operations>
98 </dsql_operation>
99 <dsql_operation operation_type="RETURN">
100 <location distribution="AllDistributions" />
101 <select>SELECT [T1_1].[col] AS [col],
102 [T1_1].[col1] AS [col1],
103 [T1_1].[StateProvinceName] AS [StateProvinceName],
104 [T1_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup],
105 [T1_1].[col2] AS [col2]
106 FROM (SELECT CONVERT (INT, [T2_1].[col], 0) AS [col],
107 CONVERT (INT, [T2_1].[col1], 0) AS [col1],
108 [T2_1].[StateProvinceName] AS [StateProvinceName],
109 [T2_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup],
110 [T2_1].[col] AS [col2]
111 FROM (SELECT CASE
112 WHEN ([T3_1].[col] = CAST ((0) AS BIGINT)) THEN CAST (NULL AS MONEY)
113 ELSE ([T3_1].[col1] / CONVERT (MONEY, [T3_1].[col], 0))
114 END AS [col],
115 CASE
116 WHEN ([T3_1].[col2] = CAST ((0) AS BIGINT)) THEN CAST (NULL AS MONEY)
117 ELSE ([T3_1].[col3] / CONVERT (MONEY, [T3_1].[col2], 0))
118 END AS [col1],
119 [T3_1].[StateProvinceName] AS [StateProvinceName],
120 [T3_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup]
121 FROM (SELECT ISNULL([T4_1].[col], CONVERT (BIGINT, 0, 0)) AS [col],
122 ISNULL([T4_1].[col1], CONVERT (MONEY, 0.00, 0)) AS [col1],
123 ISNULL([T4_1].[col2], CONVERT (BIGINT, 0, 0)) AS [col2],
124 ISNULL([T4_1].[col3], CONVERT (MONEY, 0.00, 0)) AS [col3],
125 [T4_1].[StateProvinceName] AS [StateProvinceName],
126 [T4_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup]
127 FROM (SELECT SUM([T5_1].[col]) AS [col],
128 SUM([T5_1].[col1]) AS [col1],
129 SUM([T5_1].[col2]) AS [col2],
130 SUM([T5_1].[col3]) AS [col3],
131 [T5_1].[StateProvinceName] AS [StateProvinceName],
132 [T5_1].[SalesTerritoryGroup] AS [SalesTerritoryGroup]
133 FROM [tempdb].[dbo].[TEMP_ID_16894] AS T5_1
134 GROUP BY [T5_1].[StateProvinceName], [T5_1].[SalesTerritoryGroup]) AS T4_1) AS T3_1) AS T2_1) AS T1_1
135 ORDER BY [T1_1].[col2] DESC</select>
136 </dsql_operation>
137 <dsql_operation operation_type="ON">
138 <location permanent="false" distribution="AllDistributions" />
139 <sql_operations>
140 <sql_operation type="statement">DROP TABLE [tempdb].[dbo].[TEMP_ID_16894]</sql_operation>
141 </sql_operations>
142 </dsql_operation>
143 </dsql_operations>
144 </dsql_query>
Meaning of the EXPLAIN output
The output above contains 144 numbered lines. Your output from this query may differ slightly. The following list describes significant sections.
Lines 3 through 16 provide a description of the query that is being analyzed.
Line 17, specifies that the total number of operations will be 9. You can find the start of each operation, by looking for the words dsql_operation.
Line 18 starts operation 1. Lines 18 and 19 indicate that a RND_ID operation will create a random ID number that will be used for an object description. The object described in the output above is TEMP_ID_16893. Your number will be different.
Line 20 starts operation 2. Lines 21 through 25: On all compute nodes, create a temporary table named TEMP_ID_16893.
Line 26 starts operation 3. Lines 27 through 37: Move data to TEMP_ID_16893 by using a broadcast move. The query sent to each compute node is provided. Line 37 specifies the destination table is TEMP_ID_16893.
Line 38 starts operation 4. Lines 39 through 40: Create a random ID for a table. TEMP_ID_16894 is the ID number in the example above. Your number will be different.
Line 41 starts operation 5. Lines 42 through 46: On all nodes, create a temporary table named TEMP_ID_16894.
Line 47 starts operation 6. Lines 48 through 91: Move data from various tables (including TEMP_ID_16893) to table TEMP_ID_16894, by using a shuffle move operation. The query sent to each compute node is provided. Line 90 specifies the destination table as TEMP_ID_16894. Line 91 specifies the columns.
Line 92 starts operation 7. Lines 93 through 97: On all compute nodes, drop temporary table TEMP_ID_16893.
Line 98 starts operation 8. Lines 99 through 135: Return results to the client. Uses the query provided to get the results.
Line 136 starts operation 9. Lines 137 through 140: On all nodes, drop temporary table TEMP_ID_16894.