CREATE MATERIALIZED VIEW AS

Syntax

CREATE MATERIALIZED VIEW
  materialized_view_name
  [WITH (materialized_view_parameter = value [, ... ])]
AS select_statement;

Description

CREATE MATERIALIZED VIEW AS is a statement that:

• Generates a DDL statement to create a new Materialized View.

• Launches a new query to write the results of the SELECT statement into the newly created Materialized View.

The specified select_statement generates records with columns of a primitive data type. See SELECT for the statement’s parameters and properties.

If the result of the statement is a Stream, new records are added to the Materialized View in the append mode, where new records are appended and there is no update of the existing records. Otherwise, if the statement creates a Changelog, new records are added to the Materialized View in the upsert mode, where a new record updates an existing record if there is one with matching value(s) on the PRIMARY KEY column(s). See CREATE CHANGELOG for further information on PRIMARY KEY column(s).

How to control Materialized View's size?

When creating a Materialized View, you can define a retention period for the rows to determine how long the Materialized View retains a row before deleting it. By default, a Materialized View keeps all the rows and never deletes any of them. Therefore, size of the Materialized View keeps growing as new rows are added. Defining an appropriate retention period for the Materialized View allows you to control its size and let the older rows be purged automatically. Once a row is expired (according to the retention period) and purged, it is deleted forever and its purging can not be reverted. Currently, retention period has to be defined at the creation time for a Materialized View and user can not enable/disable purging, or change the retention period later. See Materialized View Parameters to see how retention period can be defined.

Arguments

materialized_view_name

This specifies the name of the new Materialized View. Optionally, use <database_name>.<schema_name> as the prefix to the name to create the Relation in that scope.

WITH (<materialized_view_parameter> = <value> [, …​ ])

Optionally, this clause specifies Materialized View Parameters.

select_statement

This statement specifies the SELECT statement to run.

Materialized View Parameters

Parameter Name	Description
retention.millis	The duration (in milliseconds) to determine how long Materialized View retains rows before they are purged. If not set, the rows will not be purged. Required: No Default value: Infinity Type: Long
timestamp	Name of the column in the Materialized View to use as the timestamp. If not set, the timestamp of the message, coming from (one of) the Materialized view's sources, is used as the timestamp. Required: No Default value: Record’s timestamp. Type: String Valid values: Must be of type BIGINT or TIMESTAMP. See Data Types.

Examples

For the following examples, assume a Stream named pageviews has been created using the below DDL statement:

CREATE STREAM pageviews (
  viewtime BIGINT,
  userid VARCHAR,
  pageid VARCHAR
) WITH (
  'topic' = 'pageviews',
  'value.format' = 'json'
);

SELECT with a filter

The following statement creates a Materialized View, named visits, by selecting userid and pageid columns from all records in the pageviews Stream except those for User_5. Note that the result of the SELECT statement is a Stream. Therefore, new rows are simply added to the Materialized View with no impact on the existing ones.

CREATE MATERIALIZED VIEW
  visits
AS SELECT
  userid, pageid
FROM pageviews
WHERE userid <> 'User_5';

The visits Materialized View will have two columns, userid and pageid, both with the VARCHAR data type. For the details on how this view can be queried, check SELECT (FROM MATERIALIZED VIEW).

Run grouping and aggregation

The following statement creates a Materialized View, named pagevisits, by grouping records from the pageviews Stream using the pageid column and reporting the total number of records in each group as cnt. The result of the SELECT statement is a Changelog with the pageid column as the Primary Key. Therefore, new rows are added to the Materialized View in the upsert mode. Each pageid value has only one row in pagevists and a new row with matching pageid value overwrites the existing one, if any.

CREATE MATERIALIZED VIEW
  pagevisits
AS SELECT
  pageid, count(userid) as cnt 
FROM pageviews
GROUP BY pageid;

The pagevisits Materialized View will have two columns, pageid with the VARCHAR data type and cnt with the BIGINT data type. For the details on how this view can be queried, check SELECT (FROM MATERIALIZED VIEW).

Run grouping and aggregation with window function

The following statement creates a Materialized View, named visit_stats, by aggregating the total number of pages a given user visited every 30 seconds. Given that the SELECT statement generates results as a Changelog with the Primary Key of (window_start, window_end, userid), new rows are added to the Materialized View in the upsert mode, and there is only one row for any given userid in each 30-second interval.

CREATE MATERIALIZED VIEW
  visit_stats
AS SELECT 
  window_start, 
  window_end,
  userid, 
  COUNT(pageid) AS pgcnt 
FROM TUMBLE(pageviews, size 30 SECONDS) 
GROUP BY window_start, window_end, userid;

The visit_stats Materialized View will have four columns, window_start and window_end with the TIMESTAMP(3) data type, as well as userid with the VARCHAR data type and pgcnt with the BIGINT data type. For the details on how this view can be queried, check SELECT (FROM MATERIALIZED VIEW).

Create a Materialized View with retention period

The following statement creates a Materialized View named visits_rate, by aggregating the total number of pages a given user visits every 10 seconds. Moreover, the retention.millis parameter is set to 600000 milliseconds (i.e. 10 minutes). This means the retention period for any row is 10 minutes and the row is purged after that. This will help controlling the total size of visits_rate by deleting older rows which are no longer needed.

CREATE MATERIALIZED VIEW
  visits_rate
WITH ('retention.millis' = '600000')
AS SELECT 
  window_start AS start_time, 
  window_end AS end_time, 
  userid, 
  count(pageid) AS cnt
FROM Tumble(pageviews, size 10 seconds)
GROUP BY userid, window_start, window_end;