1EXPLAIN(7) PostgreSQL 15.4 Documentation EXPLAIN(7)
2
6 EXPLAIN - show the execution plan of a statement
7
9 EXPLAIN [ ( option [, ...] ) ] statement
10 EXPLAIN [ ANALYZE ] [ VERBOSE ] statement
11 where option can be one of:
13 ANALYZE [ boolean ]
15 VERBOSE [ boolean ]
16 COSTS [ boolean ]
17 SETTINGS [ boolean ]
18 BUFFERS [ boolean ]
19 WAL [ boolean ]
20 TIMING [ boolean ]
21 SUMMARY [ boolean ]
22 FORMAT { TEXT | XML | JSON | YAML }
23
25 This command displays the execution plan that the PostgreSQL planner
26 generates for the supplied statement. The execution plan shows how the
27 table(s) referenced by the statement will be scanned — by plain
28 sequential scan, index scan, etc. — and if multiple tables are
29 referenced, what join algorithms will be used to bring together the
30 required rows from each input table.
31 The most critical part of the display is the estimated statement
33 execution cost, which is the planner's guess at how long it will take
34 to run the statement (measured in cost units that are arbitrary, but
35 conventionally mean disk page fetches). Actually two numbers are shown:
36 the start-up cost before the first row can be returned, and the total
37 cost to return all the rows. For most queries the total cost is what
38 matters, but in contexts such as a subquery in EXISTS, the planner will
39 choose the smallest start-up cost instead of the smallest total cost
40 (since the executor will stop after getting one row, anyway). Also, if
41 you limit the number of rows to return with a LIMIT clause, the planner
42 makes an appropriate interpolation between the endpoint costs to
43 estimate which plan is really the cheapest.
44 The ANALYZE option causes the statement to be actually executed, not
46 only planned. Then actual run time statistics are added to the display,
47 including the total elapsed time expended within each plan node (in
48 milliseconds) and the total number of rows it actually returned. This
49 is useful for seeing whether the planner's estimates are close to
50 reality.
51 Important
53 Keep in mind that the statement is actually executed when the
54 ANALYZE option is used. Although EXPLAIN will discard any output
55 that a SELECT would return, other side effects of the statement
56 will happen as usual. If you wish to use EXPLAIN ANALYZE on an
57 INSERT, UPDATE, DELETE, MERGE, CREATE TABLE AS, or EXECUTE
58 statement without letting the command affect your data, use this
59 approach:
60 BEGIN;
62 EXPLAIN ANALYZE ...;
63 ROLLBACK;
64 Only the ANALYZE and VERBOSE options can be specified, and only in that
66 order, without surrounding the option list in parentheses. Prior to
67 PostgreSQL 9.0, the unparenthesized syntax was the only one supported.
68 It is expected that all new options will be supported only in the
69 parenthesized syntax.
70
72 ANALYZE
73 Carry out the command and show actual run times and other
74 statistics. This parameter defaults to FALSE.
75 VERBOSE
77 Display additional information regarding the plan. Specifically,
78 include the output column list for each node in the plan tree,
79 schema-qualify table and function names, always label variables in
80 expressions with their range table alias, and always print the name
81 of each trigger for which statistics are displayed. The query
82 identifier will also be displayed if one has been computed, see
83 compute_query_id for more details. This parameter defaults to
84 FALSE.
85 COSTS
87 Include information on the estimated startup and total cost of each
88 plan node, as well as the estimated number of rows and the
89 estimated width of each row. This parameter defaults to TRUE.
90 SETTINGS
92 Include information on configuration parameters. Specifically,
93 include options affecting query planning with value different from
94 the built-in default value. This parameter defaults to FALSE.
95 BUFFERS
97 Include information on buffer usage. Specifically, include the
98 number of shared blocks hit, read, dirtied, and written, the number
99 of local blocks hit, read, dirtied, and written, the number of temp
100 blocks read and written, and the time spent reading and writing
101 data file blocks and temporary file blocks (in milliseconds) if
102 track_io_timing is enabled. A hit means that a read was avoided
103 because the block was found already in cache when needed. Shared
104 blocks contain data from regular tables and indexes; local blocks
105 contain data from temporary tables and indexes; while temporary
106 blocks contain short-term working data used in sorts, hashes,
107 Materialize plan nodes, and similar cases. The number of blocks
108 dirtied indicates the number of previously unmodified blocks that
109 were changed by this query; while the number of blocks written
110 indicates the number of previously-dirtied blocks evicted from
111 cache by this backend during query processing. The number of blocks
112 shown for an upper-level node includes those used by all its child
113 nodes. In text format, only non-zero values are printed. It
114 defaults to FALSE.
115 WAL
117 Include information on WAL record generation. Specifically, include
118 the number of records, number of full page images (fpi) and the
119 amount of WAL generated in bytes. In text format, only non-zero
120 values are printed. This parameter may only be used when ANALYZE is
121 also enabled. It defaults to FALSE.
122 TIMING
124 Include actual startup time and time spent in each node in the
125 output. The overhead of repeatedly reading the system clock can
126 slow down the query significantly on some systems, so it may be
127 useful to set this parameter to FALSE when only actual row counts,
128 and not exact times, are needed. Run time of the entire statement
129 is always measured, even when node-level timing is turned off with
130 this option. This parameter may only be used when ANALYZE is also
131 enabled. It defaults to TRUE.
132 SUMMARY
134 Include summary information (e.g., totaled timing information)
135 after the query plan. Summary information is included by default
136 when ANALYZE is used but otherwise is not included by default, but
137 can be enabled using this option. Planning time in EXPLAIN EXECUTE
138 includes the time required to fetch the plan from the cache and the
139 time required for re-planning, if necessary.
140 FORMAT
142 Specify the output format, which can be TEXT, XML, JSON, or YAML.
143 Non-text output contains the same information as the text output
144 format, but is easier for programs to parse. This parameter
145 defaults to TEXT.
146 boolean
148 Specifies whether the selected option should be turned on or off.
149 You can write TRUE, ON, or 1 to enable the option, and FALSE, OFF,
150 or 0 to disable it. The boolean value can also be omitted, in which
151 case TRUE is assumed.
152 statement
154 Any SELECT, INSERT, UPDATE, DELETE, MERGE, VALUES, EXECUTE,
155 DECLARE, CREATE TABLE AS, or CREATE MATERIALIZED VIEW AS statement,
156 whose execution plan you wish to see.
157
159 The command's result is a textual description of the plan selected for
160 the statement, optionally annotated with execution statistics.
161 Section 14.1 describes the information provided.
162
164 In order to allow the PostgreSQL query planner to make reasonably
165 informed decisions when optimizing queries, the pg_statistic data
166 should be up-to-date for all tables used in the query. Normally the
167 autovacuum daemon will take care of that automatically. But if a table
168 has recently had substantial changes in its contents, you might need to
169 do a manual ANALYZE rather than wait for autovacuum to catch up with
170 the changes.
171 In order to measure the run-time cost of each node in the execution
173 plan, the current implementation of EXPLAIN ANALYZE adds profiling
174 overhead to query execution. As a result, running EXPLAIN ANALYZE on a
175 query can sometimes take significantly longer than executing the query
176 normally. The amount of overhead depends on the nature of the query, as
177 well as the platform being used. The worst case occurs for plan nodes
178 that in themselves require very little time per execution, and on
179 machines that have relatively slow operating system calls for obtaining
180 the time of day.
181
183 To show the plan for a simple query on a table with a single integer
184 column and 10000 rows:
185 EXPLAIN SELECT * FROM foo;
187 QUERY PLAN
189 ---------------------------------------------------------
190 Seq Scan on foo (cost=0.00..155.00 rows=10000 width=4)
191 (1 row)
192 Here is the same query, with JSON output formatting:
194 EXPLAIN (FORMAT JSON) SELECT * FROM foo;
196 QUERY PLAN
197 --------------------------------
198 [ +
199 { +
200 "Plan": { +
201 "Node Type": "Seq Scan",+
202 "Relation Name": "foo", +
203 "Alias": "foo", +
204 "Startup Cost": 0.00, +
205 "Total Cost": 155.00, +
206 "Plan Rows": 10000, +
207 "Plan Width": 4 +
208 } +
209 } +
210 ]
211 (1 row)
212 If there is an index and we use a query with an indexable WHERE
214 condition, EXPLAIN might show a different plan:
215 EXPLAIN SELECT * FROM foo WHERE i = 4;
217 QUERY PLAN
219 --------------------------------------------------------------
220 Index Scan using fi on foo (cost=0.00..5.98 rows=1 width=4)
221 Index Cond: (i = 4)
222 (2 rows)
223 Here is the same query, but in YAML format:
225 EXPLAIN (FORMAT YAML) SELECT * FROM foo WHERE i='4';
227 QUERY PLAN
228 -------------------------------
229 - Plan: +
230 Node Type: "Index Scan" +
231 Scan Direction: "Forward"+
232 Index Name: "fi" +
233 Relation Name: "foo" +
234 Alias: "foo" +
235 Startup Cost: 0.00 +
236 Total Cost: 5.98 +
237 Plan Rows: 1 +
238 Plan Width: 4 +
239 Index Cond: "(i = 4)"
240 (1 row)
241 XML format is left as an exercise for the reader.
243 Here is the same plan with cost estimates suppressed:
245 EXPLAIN (COSTS FALSE) SELECT * FROM foo WHERE i = 4;
247 QUERY PLAN
249 ----------------------------
250 Index Scan using fi on foo
251 Index Cond: (i = 4)
252 (2 rows)
253 Here is an example of a query plan for a query using an aggregate
255 function:
256 EXPLAIN SELECT sum(i) FROM foo WHERE i < 10;
258 QUERY PLAN
260 ---------------------------------------------------------------------
261 Aggregate (cost=23.93..23.93 rows=1 width=4)
262 -> Index Scan using fi on foo (cost=0.00..23.92 rows=6 width=4)
263 Index Cond: (i < 10)
264 (3 rows)
265 Here is an example of using EXPLAIN EXECUTE to display the execution
267 plan for a prepared query:
268 PREPARE query(int, int) AS SELECT sum(bar) FROM test
270 WHERE id > $1 AND id < $2
271 GROUP BY foo;
272 EXPLAIN ANALYZE EXECUTE query(100, 200);
274 QUERY PLAN
276 ------------------------------------------------------------------------------------------------------------------------
277 HashAggregate (cost=9.54..9.54 rows=1 width=8) (actual time=0.156..0.161 rows=11 loops=1)
278 Group Key: foo
279 -> Index Scan using test_pkey on test (cost=0.29..9.29 rows=50 width=8) (actual time=0.039..0.091 rows=99 loops=1)
280 Index Cond: ((id > $1) AND (id < $2))
281 Planning time: 0.197 ms
282 Execution time: 0.225 ms
283 (6 rows)
284 Of course, the specific numbers shown here depend on the actual
286 contents of the tables involved. Also note that the numbers, and even
287 the selected query strategy, might vary between PostgreSQL releases due
288 to planner improvements. In addition, the ANALYZE command uses random
289 sampling to estimate data statistics; therefore, it is possible for
290 cost estimates to change after a fresh run of ANALYZE, even if the
291 actual distribution of data in the table has not changed.
292
294 There is no EXPLAIN statement defined in the SQL standard.
295
297 ANALYZE(7)
298PostgreSQL 15.4 2023 EXPLAIN(7)