1CREATE INDEX(7) PostgreSQL 11.6 Documentation CREATE INDEX(7)
2
6 CREATE_INDEX - define a new index
7
9 CREATE [ UNIQUE ] INDEX [ CONCURRENTLY ] [ [ IF NOT EXISTS ] name ] ON [ ONLY ] table_name [ USING method ]
10 ( { column_name | ( expression ) } [ COLLATE collation ] [ opclass ] [ ASC | DESC ] [ NULLS { FIRST | LAST } ] [, ...] )
11 [ INCLUDE ( column_name [, ...] ) ]
12 [ WITH ( storage_parameter = value [, ... ] ) ]
13 [ TABLESPACE tablespace_name ]
14 [ WHERE predicate ]
15
17 CREATE INDEX constructs an index on the specified column(s) of the
18 specified relation, which can be a table or a materialized view.
19 Indexes are primarily used to enhance database performance (though
20 inappropriate use can result in slower performance).
21 The key field(s) for the index are specified as column names, or
23 alternatively as expressions written in parentheses. Multiple fields
24 can be specified if the index method supports multicolumn indexes.
25 An index field can be an expression computed from the values of one or
27 more columns of the table row. This feature can be used to obtain fast
28 access to data based on some transformation of the basic data. For
29 example, an index computed on upper(col) would allow the clause WHERE
30 upper(col) = 'JIM' to use an index.
31 PostgreSQL provides the index methods B-tree, hash, GiST, SP-GiST, GIN,
33 and BRIN. Users can also define their own index methods, but that is
34 fairly complicated.
35 When the WHERE clause is present, a partial index is created. A partial
37 index is an index that contains entries for only a portion of a table,
38 usually a portion that is more useful for indexing than the rest of the
39 table. For example, if you have a table that contains both billed and
40 unbilled orders where the unbilled orders take up a small fraction of
41 the total table and yet that is an often used section, you can improve
42 performance by creating an index on just that portion. Another possible
43 application is to use WHERE with UNIQUE to enforce uniqueness over a
44 subset of a table. See Section 11.8 for more discussion.
45 The expression used in the WHERE clause can refer only to columns of
47 the underlying table, but it can use all columns, not just the ones
48 being indexed. Presently, subqueries and aggregate expressions are also
49 forbidden in WHERE. The same restrictions apply to index fields that
50 are expressions.
51 All functions and operators used in an index definition must be
53 “immutable”, that is, their results must depend only on their arguments
54 and never on any outside influence (such as the contents of another
55 table or the current time). This restriction ensures that the behavior
56 of the index is well-defined. To use a user-defined function in an
57 index expression or WHERE clause, remember to mark the function
58 immutable when you create it.
59
61 UNIQUE
62 Causes the system to check for duplicate values in the table when
63 the index is created (if data already exist) and each time data is
64 added. Attempts to insert or update data which would result in
65 duplicate entries will generate an error.
66 Additional restrictions apply when unique indexes are applied to
68 partitioned tables; see CREATE TABLE (CREATE_TABLE(7)).
69 CONCURRENTLY
71 When this option is used, PostgreSQL will build the index without
72 taking any locks that prevent concurrent inserts, updates, or
73 deletes on the table; whereas a standard index build locks out
74 writes (but not reads) on the table until it's done. There are
75 several caveats to be aware of when using this option — see
76 Building Indexes Concurrently.
77 IF NOT EXISTS
79 Do not throw an error if a relation with the same name already
80 exists. A notice is issued in this case. Note that there is no
81 guarantee that the existing index is anything like the one that
82 would have been created. Index name is required when IF NOT EXISTS
83 is specified.
84 INCLUDE
86 The optional INCLUDE clause specifies a list of columns which will
87 be included in the index as non-key columns. A non-key column
88 cannot be used in an index scan search qualification, and it is
89 disregarded for purposes of any uniqueness or exclusion constraint
90 enforced by the index. However, an index-only scan can return the
91 contents of non-key columns without having to visit the index's
92 table, since they are available directly from the index entry.
93 Thus, addition of non-key columns allows index-only scans to be
94 used for queries that otherwise could not use them.
95 It's wise to be conservative about adding non-key columns to an
97 index, especially wide columns. If an index tuple exceeds the
98 maximum size allowed for the index type, data insertion will fail.
99 In any case, non-key columns duplicate data from the index's table
100 and bloat the size of the index, thus potentially slowing searches.
101 Columns listed in the INCLUDE clause don't need appropriate
103 operator classes; the clause can include columns whose data types
104 don't have operator classes defined for a given access method.
105 Expressions are not supported as included columns since they cannot
107 be used in index-only scans.
108 Currently, only the B-tree index access method supports this
110 feature. In B-tree indexes, the values of columns listed in the
111 INCLUDE clause are included in leaf tuples which correspond to heap
112 tuples, but are not included in upper-level index entries used for
113 tree navigation.
114 name
116 The name of the index to be created. No schema name can be included
117 here; the index is always created in the same schema as its parent
118 table. If the name is omitted, PostgreSQL chooses a suitable name
119 based on the parent table's name and the indexed column name(s).
120 ONLY
122 Indicates not to recurse creating indexes on partitions, if the
123 table is partitioned. The default is to recurse.
124 table_name
126 The name (possibly schema-qualified) of the table to be indexed.
127 method
129 The name of the index method to be used. Choices are btree, hash,
130 gist, spgist, gin, and brin. The default method is btree.
131 column_name
133 The name of a column of the table.
134 expression
136 An expression based on one or more columns of the table. The
137 expression usually must be written with surrounding parentheses, as
138 shown in the syntax. However, the parentheses can be omitted if the
139 expression has the form of a function call.
140 collation
142 The name of the collation to use for the index. By default, the
143 index uses the collation declared for the column to be indexed or
144 the result collation of the expression to be indexed. Indexes with
145 non-default collations can be useful for queries that involve
146 expressions using non-default collations.
147 opclass
149 The name of an operator class. See below for details.
150 ASC
152 Specifies ascending sort order (which is the default).
153 DESC
155 Specifies descending sort order.
156 NULLS FIRST
158 Specifies that nulls sort before non-nulls. This is the default
159 when DESC is specified.
160 NULLS LAST
162 Specifies that nulls sort after non-nulls. This is the default when
163 DESC is not specified.
164 storage_parameter
166 The name of an index-method-specific storage parameter. See Index
167 Storage Parameters for details.
168 tablespace_name
170 The tablespace in which to create the index. If not specified,
171 default_tablespace is consulted, or temp_tablespaces for indexes on
172 temporary tables.
173 predicate
175 The constraint expression for a partial index.
176 Index Storage Parameters
178 The optional WITH clause specifies storage parameters for the index.
179 Each index method has its own set of allowed storage parameters. The
180 B-tree, hash, GiST and SP-GiST index methods all accept this parameter:
181 fillfactor
183 The fillfactor for an index is a percentage that determines how
184 full the index method will try to pack index pages. For B-trees,
185 leaf pages are filled to this percentage during initial index
186 build, and also when extending the index at the right (adding new
187 largest key values). If pages subsequently become completely full,
188 they will be split, leading to gradual degradation in the index's
189 efficiency. B-trees use a default fillfactor of 90, but any integer
190 value from 10 to 100 can be selected. If the table is static then
191 fillfactor 100 is best to minimize the index's physical size, but
192 for heavily updated tables a smaller fillfactor is better to
193 minimize the need for page splits. The other index methods use
194 fillfactor in different but roughly analogous ways; the default
195 fillfactor varies between methods.
196 B-tree indexes additionally accept this parameter:
198 vacuum_cleanup_index_scale_factor
200 Per-index value for vacuum_cleanup_index_scale_factor.
201 GiST indexes additionally accept this parameter:
203 buffering
205 Determines whether the buffering build technique described in
206 Section 64.4.1 is used to build the index. With OFF it is disabled,
207 with ON it is enabled, and with AUTO it is initially disabled, but
208 turned on on-the-fly once the index size reaches
209 effective_cache_size. The default is AUTO.
210 GIN indexes accept different parameters:
212 fastupdate
214 This setting controls usage of the fast update technique described
215 in Section 66.4.1. It is a Boolean parameter: ON enables fast
216 update, OFF disables it. (Alternative spellings of ON and OFF are
217 allowed as described in Section 19.1.) The default is ON.
218 Note
220 Turning fastupdate off via ALTER INDEX prevents future
221 insertions from going into the list of pending index entries,
222 but does not in itself flush previous entries. You might want
223 to VACUUM the table or call gin_clean_pending_list function
224 afterward to ensure the pending list is emptied.
225 gin_pending_list_limit
227 Custom gin_pending_list_limit parameter. This value is specified in
228 kilobytes.
229 BRIN indexes accept different parameters:
231 pages_per_range
233 Defines the number of table blocks that make up one block range for
234 each entry of a BRIN index (see Section 67.1 for more details). The
235 default is 128.
236 autosummarize
238 Defines whether a summarization run is invoked for the previous
239 page range whenever an insertion is detected on the next one.
240 Building Indexes Concurrently
242 Creating an index can interfere with regular operation of a database.
243 Normally PostgreSQL locks the table to be indexed against writes and
244 performs the entire index build with a single scan of the table. Other
245 transactions can still read the table, but if they try to insert,
246 update, or delete rows in the table they will block until the index
247 build is finished. This could have a severe effect if the system is a
248 live production database. Very large tables can take many hours to be
249 indexed, and even for smaller tables, an index build can lock out
250 writers for periods that are unacceptably long for a production system.
251 PostgreSQL supports building indexes without locking out writes. This
253 method is invoked by specifying the CONCURRENTLY option of CREATE
254 INDEX. When this option is used, PostgreSQL must perform two scans of
255 the table, and in addition it must wait for all existing transactions
256 that could potentially modify or use the index to terminate. Thus this
257 method requires more total work than a standard index build and takes
258 significantly longer to complete. However, since it allows normal
259 operations to continue while the index is built, this method is useful
260 for adding new indexes in a production environment. Of course, the
261 extra CPU and I/O load imposed by the index creation might slow other
262 operations.
263 In a concurrent index build, the index is actually entered into the
265 system catalogs in one transaction, then two table scans occur in two
266 more transactions. Before each table scan, the index build must wait
267 for existing transactions that have modified the table to terminate.
268 After the second scan, the index build must wait for any transactions
269 that have a snapshot (see Chapter 13) predating the second scan to
270 terminate. Then finally the index can be marked ready for use, and the
271 CREATE INDEX command terminates. Even then, however, the index may not
272 be immediately usable for queries: in the worst case, it cannot be used
273 as long as transactions exist that predate the start of the index
274 build.
275 If a problem arises while scanning the table, such as a deadlock or a
277 uniqueness violation in a unique index, the CREATE INDEX command will
278 fail but leave behind an “invalid” index. This index will be ignored
279 for querying purposes because it might be incomplete; however it will
280 still consume update overhead. The psql \d command will report such an
281 index as INVALID:
282 postgres=# \d tab
284 Table "public.tab"
285 Column | Type | Collation | Nullable | Default
286 --------+---------+-----------+----------+---------
287 col | integer | | |
288 Indexes:
289 "idx" btree (col) INVALID
290 The recommended recovery method in such cases is to drop the index and
292 try again to perform CREATE INDEX CONCURRENTLY. (Another possibility is
293 to rebuild the index with REINDEX. However, since REINDEX does not
294 support concurrent builds, this option is unlikely to seem attractive.)
295 Another caveat when building a unique index concurrently is that the
297 uniqueness constraint is already being enforced against other
298 transactions when the second table scan begins. This means that
299 constraint violations could be reported in other queries prior to the
300 index becoming available for use, or even in cases where the index
301 build eventually fails. Also, if a failure does occur in the second
302 scan, the “invalid” index continues to enforce its uniqueness
303 constraint afterwards.
304 Concurrent builds of expression indexes and partial indexes are
306 supported. Errors occurring in the evaluation of these expressions
307 could cause behavior similar to that described above for unique
308 constraint violations.
309 Regular index builds permit other regular index builds on the same
311 table to occur simultaneously, but only one concurrent index build can
312 occur on a table at a time. In either case, schema modification of the
313 table is not allowed while the index is being built. Another difference
314 is that a regular CREATE INDEX command can be performed within a
315 transaction block, but CREATE INDEX CONCURRENTLY cannot.
316 Concurrent builds for indexes on partitioned tables are currently not
318 supported. However, you may concurrently build the index on each
319 partition individually and then finally create the partitioned index
320 non-concurrently in order to reduce the time where writes to the
321 partitioned table will be locked out. In this case, building the
322 partitioned index is a metadata only operation.
323
325 See Chapter 11 for information about when indexes can be used, when
326 they are not used, and in which particular situations they can be
327 useful.
328 Currently, only the B-tree, GiST, GIN, and BRIN index methods support
330 multicolumn indexes. Up to 32 fields can be specified by default. (This
331 limit can be altered when building PostgreSQL.) Only B-tree currently
332 supports unique indexes.
333 An operator class can be specified for each column of an index. The
335 operator class identifies the operators to be used by the index for
336 that column. For example, a B-tree index on four-byte integers would
337 use the int4_ops class; this operator class includes comparison
338 functions for four-byte integers. In practice the default operator
339 class for the column's data type is usually sufficient. The main point
340 of having operator classes is that for some data types, there could be
341 more than one meaningful ordering. For example, we might want to sort a
342 complex-number data type either by absolute value or by real part. We
343 could do this by defining two operator classes for the data type and
344 then selecting the proper class when creating an index. More
345 information about operator classes is in Section 11.10 and in
346 Section 38.15.
347 When CREATE INDEX is invoked on a partitioned table, the default
349 behavior is to recurse to all partitions to ensure they all have
350 matching indexes. Each partition is first checked to determine whether
351 an equivalent index already exists, and if so, that index will become
352 attached as a partition index to the index being created, which will
353 become its parent index. If no matching index exists, a new index will
354 be created and automatically attached; the name of the new index in
355 each partition will be determined as if no index name had been
356 specified in the command. If the ONLY option is specified, no recursion
357 is done, and the index is marked invalid. (ALTER INDEX ... ATTACH
358 PARTITION marks the index valid, once all partitions acquire matching
359 indexes.) Note, however, that any partition that is created in the
360 future using CREATE TABLE ... PARTITION OF will automatically have a
361 matching index, regardless of whether ONLY is specified.
362 For index methods that support ordered scans (currently, only B-tree),
364 the optional clauses ASC, DESC, NULLS FIRST, and/or NULLS LAST can be
365 specified to modify the sort ordering of the index. Since an ordered
366 index can be scanned either forward or backward, it is not normally
367 useful to create a single-column DESC index — that sort ordering is
368 already available with a regular index. The value of these options is
369 that multicolumn indexes can be created that match the sort ordering
370 requested by a mixed-ordering query, such as SELECT ... ORDER BY x ASC,
371 y DESC. The NULLS options are useful if you need to support “nulls sort
372 low” behavior, rather than the default “nulls sort high”, in queries
373 that depend on indexes to avoid sorting steps.
374 For most index methods, the speed of creating an index is dependent on
376 the setting of maintenance_work_mem. Larger values will reduce the time
377 needed for index creation, so long as you don't make it larger than the
378 amount of memory really available, which would drive the machine into
379 swapping.
380 PostgreSQL can build indexes while leveraging multiple CPUs in order to
382 process the table rows faster. This feature is known as parallel index
383 build. For index methods that support building indexes in parallel
384 (currently, only B-tree), maintenance_work_mem specifies the maximum
385 amount of memory that can be used by each index build operation as a
386 whole, regardless of how many worker processes were started. Generally,
387 a cost model automatically determines how many worker processes should
388 be requested, if any.
389 Parallel index builds may benefit from increasing maintenance_work_mem
391 where an equivalent serial index build will see little or no benefit.
392 Note that maintenance_work_mem may influence the number of worker
393 processes requested, since parallel workers must have at least a 32MB
394 share of the total maintenance_work_mem budget. There must also be a
395 remaining 32MB share for the leader process. Increasing
396 max_parallel_maintenance_workers may allow more workers to be used,
397 which will reduce the time needed for index creation, so long as the
398 index build is not already I/O bound. Of course, there should also be
399 sufficient CPU capacity that would otherwise lie idle.
400 Setting a value for parallel_workers via ALTER TABLE (ALTER_TABLE(7))
402 directly controls how many parallel worker processes will be requested
403 by a CREATE INDEX against the table. This bypasses the cost model
404 completely, and prevents maintenance_work_mem from affecting how many
405 parallel workers are requested. Setting parallel_workers to 0 via ALTER
406 TABLE will disable parallel index builds on the table in all cases.
407 Tip
409 You might want to reset parallel_workers after setting it as part
410 of tuning an index build. This avoids inadvertent changes to query
411 plans, since parallel_workers affects all parallel table scans.
412 While CREATE INDEX with the CONCURRENTLY option supports parallel
414 builds without special restrictions, only the first table scan is
415 actually performed in parallel.
416 Use DROP INDEX (DROP_INDEX(7)) to remove an index.
418 Prior releases of PostgreSQL also had an R-tree index method. This
420 method has been removed because it had no significant advantages over
421 the GiST method. If USING rtree is specified, CREATE INDEX will
422 interpret it as USING gist, to simplify conversion of old databases to
423 GiST.
424
426 To create a unique B-tree index on the column title in the table films:
427 CREATE UNIQUE INDEX title_idx ON films (title);
429 To create a unique B-tree index on the column title with included
431 columns director and rating in the table films:
432 CREATE UNIQUE INDEX title_idx ON films (title) INCLUDE (director, rating);
434 To create an index on the expression lower(title), allowing efficient
436 case-insensitive searches:
437 CREATE INDEX ON films ((lower(title)));
439 (In this example we have chosen to omit the index name, so the system
441 will choose a name, typically films_lower_idx.)
442 To create an index with non-default collation:
444 CREATE INDEX title_idx_german ON films (title COLLATE "de_DE");
446 To create an index with non-default sort ordering of nulls:
448 CREATE INDEX title_idx_nulls_low ON films (title NULLS FIRST);
450 To create an index with non-default fill factor:
452 CREATE UNIQUE INDEX title_idx ON films (title) WITH (fillfactor = 70);
454 To create a GIN index with fast updates disabled:
456 CREATE INDEX gin_idx ON documents_table USING GIN (locations) WITH (fastupdate = off);
458 To create an index on the column code in the table films and have the
460 index reside in the tablespace indexspace:
461 CREATE INDEX code_idx ON films (code) TABLESPACE indexspace;
463 To create a GiST index on a point attribute so that we can efficiently
465 use box operators on the result of the conversion function:
466 CREATE INDEX pointloc
468 ON points USING gist (box(location,location));
469 SELECT * FROM points
470 WHERE box(location,location) && '(0,0),(1,1)'::box;
471 To create an index without locking out writes to the table:
473 CREATE INDEX CONCURRENTLY sales_quantity_index ON sales_table (quantity);
475
477 CREATE INDEX is a PostgreSQL language extension. There are no
478 provisions for indexes in the SQL standard.
479
481 ALTER INDEX (ALTER_INDEX(7)), DROP INDEX (DROP_INDEX(7))
482PostgreSQL 11.6 2019 CREATE INDEX(7)