The pg_ivm module provides Incremental View Maintenance (IVM) feature for PostgreSQL.
The extension is compatible with PostgreSQL 13, 14, and 15.
Incremental View Maintenance (IVM) is a way to make materialized views up-to-date in which only incremental changes are computed and applied on views rather than recomputing the contents from scratch as REFRESH MATERIALIZED VIEW does. IVM can update materialized views more efficiently than recomputation when only small parts of the view are changed.
There are two approaches with regard to timing of view maintenance: immediate and deferred. In immediate maintenance, views are updated in the same transaction that its base table is modified. In deferred maintenance, views are updated after the transaction is committed, for example, when the view is accessed, as a response to user command like REFRESH MATERIALIZED VIEW, or periodically in background, and so on. pg_ivm provides a kind of immediate maintenance, in which materialized views are updated immediately in AFTER triggers when a base table is modified.
We call a materialized view supporting IVM an Incrementally Maintainable Materialized View (IMMV). To create IMMV, you have to call create_immv function with a relation name and a view definition query. For example:
SELECT create_immv('myview', 'SELECT * FROM mytab');creates an IMMV with name 'myview' defined as SELECT * FROM mytab. This is corresponding to the following command to create a normal materialized view;
CREATE MATERIALIZED VIEW myview AS SELECT * FROM mytab;When an IMMV is created, some triggers are automatically created so that the view's contents are immediately updated when its base tables are modified.
postgres=# SELECT create_immv('m', 'SELECT * FROM t0');
NOTICE: could not create an index on immv "m" automatically
DETAIL: This target list does not have all the primary key columns, or this view does not contain DISTINCT clause.
HINT: Create an index on the immv for efficient incremental maintenance.
create_immv
-------------
3
(1 row)
postgres=# SELECT * FROM m;
i
---
1
2
3
(3 rows)
postgres=# INSERT INTO t0 VALUES (4);
INSERT 0 1
postgres=# SELECT * FROM m; -- automatically updated
i
---
1
2
3
4
(4 rows)To install pg_ivm, execute this in the module's directory:
make installIf you installed PostgreSQL from rpm or deb, you will need the devel package (for example, postgresql14-devel or postgresql-server-dev-14).
Important: Don't forget to set the
PG_CONFIGvariable (make PG_CONFIG=...) or thePATHto thepg_configcommand in case you want to usepg_ivmon a non-default or custom build of PostgreSQL. Read more here.
And, execute CREATE EXTENSION comand.
CREATE EXTENSION pg_ivm;
RPM packages of pg_ivm are available from the PostgreSQL yum repository. See the instruction for details. Note that we are not the maintainer of this yum repository and RPMs for pg_ivm in it may be not always latest.
When pg_ivm is installed, the following objects are created.
Use create_immv function to create IMMV.
create_immv(immv_name text, view_definition text) RETURNS bigint
create_immv defines a new IMMV of a query. A table of the name immv_name is created and a query specified by view_definition is executed and used to populate the IMMV. The query is stored in pg_ivm_immv, so that it can be refreshed later upon incremental view maintenance. create_immv returns the number of rows in the created IMMV.
When an IMMV is created, some triggers are automatically created so that the view's contents are immediately updated when its base tables are modified. In addition, a unique index is created on the IMMV automatically if possible. If the view definition query has a GROUP BY clause, a unique index is created on the columns of GROUP BY expressions. Also, if the view has DISTINCT clause, a unique index is created on all columns in the target list. Otherwise, if the IMMV contains all primary key attritubes of its base tables in the target list, a unique index is created on these attritubes. In other cases, no index is created.
Use refresh_immv function to refresh IMMV.
refresh_immv(immv_name text, with_data bool) RETURNS bigint
refresh_immv completely replaces the contents of an IMMV as REFRESH MATERIALIZED VIEW command does for a materialized view. To execute this function you must be the owner of the IMMV. The old contents are discarded.
The with_data flag is corresponding to WITH [NO] DATA option of REFRESH MATERIALIZED VIEW` command. If with_data is true, the backing query is executed to provide the new data, and if the IMMV is unpopulated, triggers for maintaining the view are created. Also, a unique index is created for IMMV if it is possible and the view doesn't have that yet. If with_data is false, no new data is generated and the IMMV become unpopulated, and the triggers are dropped from the IMMV. Note that unpopulated IMMV is still scannable although the result is empty. This behaviour may be changed in future to raise an error when an unpopulated IMMV is scanned.
The catalog pg_ivm_immv stores IMMV information.
| Name | Type | Description |
|---|---|---|
| immvrelid | regclass | The OID of the IMMV |
| viewdef | text | Query tree (in the form of a nodeToString() representation) for the view definition |
| ispopulated | bool | True if IMMV is currently populated |
In general, IMMVs allow faster updates than REFRESH MATERIALIZED VIEW at the price of slower updates to their base tables. Update of base tables is slower because triggers will be invoked and the IMMV is updated in triggers per modification statement.
For example, suppose a normal materialized view defined as below:
test=# CREATE MATERIALIZED VIEW mv_normal AS
SELECT a.aid, b.bid, a.abalance, b.bbalance
FROM pgbench_accounts a JOIN pgbench_branches b USING(bid);
SELECT 10000000Updating a tuple in a base table of this materialized view is rapid but the REFRESH MATERIALIZED VIEW command on this view takes a long time:
test=# UPDATE pgbench_accounts SET abalance = 1000 WHERE aid = 1;
UPDATE 1
Time: 9.052 ms
test=# REFRESH MATERIALIZED VIEW mv_normal ;
REFRESH MATERIALIZED VIEW
Time: 20575.721 ms (00:20.576)On the other hand, after creating IMMV with the same view definition as below:
test=# SELECT create_immv('immv',
'SELECT a.aid, b.bid, a.abalance, b.bbalance
FROM pgbench_accounts a JOIN pgbench_branches b USING(bid)');
NOTICE: created index "immv_index" on immv "immv"
create_immv
-------------
10000000
(1 row)
updating a tuple in a base table takes more than the normal view, but its content is updated automatically and this is faster than the REFRESH MATERIALIZED VIEW command.
test=# UPDATE pgbench_accounts SET abalance = 1234 WHERE aid = 1;
UPDATE 1
Time: 15.448 ms
test=# SELECT * FROM immv WHERE aid = 1;
aid | bid | abalance | bbalance
-----+-----+----------+----------
1 | 1 | 1234 | 0
(1 row)An appropriate index on IMMV is necessary for efficient IVM because we need to looks for tuples to be updated in IMMV. If there are no indexes, it will take a long time.
Therefore, when an IMMV is created by the create_immv function, a unique index is created on it automatically if possible. If the view definition query has a GROUP BY clause, a unique index is created on the columns of GROUP BY expressions. Also, if the view has DISTINCT clause, a unique index is created on all columns in the target list. Otherwise, if the IMMV contains all primary key attritubes of its base tables in the target list, a unique index is created on these attritubes. In other cases, no index is created.
In the previous example, a unique index "immv_index" is created on aid and bid columns of "immv", and this enables the rapid update of the view. Dropping this index make updating the view take a loger time.
test=# DROP INDEX immv_index;
DROP INDEX
test=# UPDATE pgbench_accounts SET abalance = 9876 WHERE aid = 1;
UPDATE 1
Time: 3224.741 ms (00:03.225)Currently, IMMV's view definition can contain inner joins, DISTINCT clause, some built-in aggregate functions, and simple sub-queries in FROM clause. Inner joins including self-join are supported, but outer joins are not supported. Supported aggregate functions are count, sum, avg, min and max. Other aggregates, sub-queries which contain an aggregate or DISTINCT clause, sub-queries in other than FROM clause, CTEs, window functions, HAVING, ORDER BY, LIMIT/OFFSET, UNION/INTERSECT/EXCEPT, DISTINCT ON, TABLESAMPLE, VALUES, and FOR UPDATE/SHARE can not be used in view definition.
The base tables must be simple tables. Views, materialized views, inheritance parent tables, partitioned tables, partitions, and foreign tables can not be used.
The targetlist cannot contain system columns, columns whose name starts with __ivm_.
Logical replication is not supported, that is, even when a base table at a publisher node is modified, IMMVs at subscriber nodes defined on these base tables are not updated.
Supported aggregate functions are count, sum, avg, min, and max. Currently, only built-in aggregate functions are supported and user defined aggregates cannot be used.
When an IMMV including aggregate is created, some extra columns whose name start with __ivm are automatically added to the target list. __ivm_count__ contains the number of tuples aggregated in each group. In addition, more than one extra columns for each column of aggregated value are added in order to maintain the value. For example, columns named like __ivm_count_avg__ and __ivm_sum_avg__ are added for maintaining an average value. When a base table is modified, the new aggregated values are incrementally calculated using the old aggregated values and values of related extra columns stored in the IMMV.
Note that for min or max, the new values could be re-calculated from base tables with regard to the affected groups when a tuple containing the current minimal or maximal values are deleted from a base table. Therefore, it can takes a long time to update an IMMV containing these functions.
Also, note that using sum or avg on real (float4) type or double precision (float8) type in IMMV is unsafe, because aggregated values in IMMV can become different from results calculated from base tables due to the limited precision of these types. To avoid this problem, use the numeric type instead.
If we have a GROUP BY clause, expressions specified in GROUP BY must appear in the target list. This is how tuples to be updated in the IMMV are identified. These attributes are used as scan keys for searching tuples in the IMMV, so indexes on them are required for efficient IVM.
Targetlist cannot contain expressions which contain an aggregate in it.
Simple subqueries in FROM clause are supported.
Subqueries can be used only in FROM clause. Subqueries in target list or WHERE clause are not supported.
Subqueries containing an aggregate function or DISTINCT are not supported.
DISTINCT is allowed in IMMV's definition queries. Suppose an IMMV defined with DISTINCT on a base table containing duplicate tuples. When tuples are deleted from the base table, a tuple in the view is deleted if and only if the multiplicity of the tuple becomes zero. Moreover, when tuples are inserted into the base table, a tuple is inserted into the view only if the same tuple doesn't already exist in it.
Physically, an IMMV defined with DISTINCT contains tuples after eliminating duplicates, and the multiplicity of each tuple is stored in a extra column named __ivm_count__ that is added when such IMMV is created.
When a base table is truncated, the IMMV is also truncated and the contents become empty if the view definition query does not contain an aggregate without a GROUP BY clause. Aggregate views without a GROUP BY clause always have one row. Therefore, in such cases, if a base table is truncated, the IMMV is simply refreshed instead of being truncated.
Suppose an IMMV is defined on two base tables and each table was modified in different a concurrent transaction simultaneously. In the transaction which was committed first, the IMMV can be updated considering only the change which happened in this transaction. On the other hand, in order to update the IMMV correctly in the transaction which was committed later, we need to know the changes occurred in both transactions. For this reason, ExclusiveLock is held on an IMMV immediately after a base table is modified in READ COMMITTED mode to make sure that the IMMV is updated in the latter transaction after the former transaction is committed. In REPEATABLE READ or SERIALIZABLE mode, an error is raised immediately if lock acquisition fails because any changes which occurred in other transactions are not be visible in these modes and IMMV cannot be updated correctly in such situations. However, as an exception if the IMMV has only one base table and doesn't use DISTINCT or GROUP BY, the lock held on the IMMV is RowExclusiveLock.
If some base tables have row level security policy, rows that are not visible to the materialized view's owner are excluded from the result. In addition, such rows are excluded as well when views are incrementally maintained. However, if a new policy is defined or policies are changed after the materialized view was created, the new policy will not be applied to the view contents. To apply the new policy, you need to recreate IMMV.
IVM is effective when we want to keep an IMMV up-to-date and small fraction of a base table is modified infrequently. Due to the overhead of immediate maintenance, IVM is not effective when a base table is modified frequently. Also, when a large part of a base table is modified or large data is inserted into a base table, IVM is not effective and the cost of maintenance can be larger than refresh from scratch.
In such situation, we can use refesh_immv function with with_data = falase to disable immediate maintenance before modifying a base table. After a base table modification, call refresh_immvwith with_data = true to refresh the view data and enable immediate maintenance.
IVM Development Group
- Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group
- Portions Copyright (c) 2022, IVM Development Group