Window

GROUP WINDOW

Description

Group Window is defined in GROUP BY. One record is generated from each group. Group Window involves the following functions:

• Array functions

  Table 1 Array functions

  Grouping Window Function	Description
  TUMBLE(time_attr, interval)	Defines a tumbling time window. A tumbling time window assigns rows to non-overlapping, continuous windows with a fixed duration (interval). For example, a tumbling window of 5 minutes groups rows in 5 minutes intervals. Tumbling windows can be defined on event-time (stream + batch) or processing-time (stream).
  HOP(time_attr, interval, interval)	Defines a hopping time window (called sliding window in the Table API). A hopping time window has a fixed duration (second interval parameter) and hops by a specified hop interval (first interval parameter). If the hop interval is smaller than the window size, hopping windows are overlapping. Thus, rows can be assigned to multiple windows. For example, a hopping window of 15 minutes size and 5 minute hop interval assigns each row to 3 different windows of 15 minute size, which are evaluated in an interval of 5 minutes. Hopping windows can be defined on event-time (stream + batch) or processing-time (stream).
  SESSION(time_attr, interval)	Defines a session time window. Session time windows do not have a fixed duration but their bounds are defined by a time interval of inactivity, i.e., a session window is closed if no event appears for a defined gap period. For example a session window with a 30 minute gap starts when a row is observed after 30 minutes inactivity (otherwise the row would be added to an existing window) and is closed if no row is added within 30 minutes. Session windows can work on event-time (stream + batch) or processing-time (stream).

  

  In streaming mode, the time_attr argument of the group window function must refer to a valid time attribute that specifies the processing time or event time of rows.

  • event-time: The type is timestamp(3).
  • processing-time: No need to specify the type.

  In batch mode, the time_attr argument of the group window function must be an attribute of type timestamp.

• Window helper functions
  You can use the following helper functions to select the start and end timestamps, as well as the time attribute, for grouping windows.

  Table 2 Window helper functions

  Helper Function	Description
  TUMBLE_START(time_attr, interval) HOP_START(time_attr, interval, interval) SESSION_START(time_attr, interval)	Returns the timestamp of the inclusive lower bound of the corresponding tumbling, hopping, or session window.
  TUMBLE_END(time_attr, interval) HOP_END(time_attr, interval, interval) SESSION_END(time_attr, interval)	Returns the timestamp of the exclusive upper bound of the corresponding tumbling, hopping, or session window. Note: The exclusive upper bound timestamp cannot be used as a rowtime attribute in subsequent time-based operations, such as interval joins and group window or over window aggregations.
  TUMBLE_ROWTIME(time_attr, interval) HOP_ROWTIME(time_attr, interval, interval) SESSION_ROWTIME(time_attr, interval)	Returns the timestamp of the inclusive upper bound of the corresponding tumbling, hopping, or session window. The resulting attribute is a rowtime attribute that can be used in subsequent time-based operations such as interval joins and group window or over window aggregations.
  TUMBLE_PROCTIME(time_attr, interval) HOP_PROCTIME(time_attr, interval, interval) SESSION_PROCTIME(time_attr, interval)	Returns a proctime attribute that can be used in subsequent time-based operations such as interval joins and group window or over window aggregations.

  Note: When calling helper functions, it is important to use the same parameters as those used in the GROUP BY clause for grouping window functions.

Example

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

// Calculate the SUM every day (event time).
insert into temp SELECT name,
    TUMBLE_START(ts, INTERVAL '1' DAY) as wStart,
    SUM(amount)
    FROM Orders
    GROUP BY TUMBLE(ts, INTERVAL '1' DAY), name;

// Calculate the SUM every day (processing time). 
insert into temp SELECT name, 
    SUM(amount) 
    FROM Orders 
    GROUP BY TUMBLE(proctime, INTERVAL '1' DAY), name;

// Calculate the SUM over the recent 24 hours every hour (event time).
insert into temp SELECT product, 
    SUM(amount) 
    FROM Orders 
    GROUP BY HOP(ts, INTERVAL '1' HOUR, INTERVAL '1' DAY), product;

// Calculate the SUM of each session and an inactive interval every 12 hours (event time).
insert into temp SELECT name, 
    SESSION_START(ts, INTERVAL '12' HOUR) AS sStart,
    SESSION_END(ts, INTERVAL '12' HOUR) AS sEnd,
    SUM(amount)
    FROM Orders
    GROUP BY SESSION(ts, INTERVAL '12' HOUR), name;

TUMBLE WINDOW Extension

Function

Precautions

• If you use the INSERT statement to write results to a sink, it must support the upsert mode. Ensure that the result table supports upsert operations and the primary key is defined.
• Latency settings only take effect for event time and not for proctime.
• When calling helper functions, it is important to use the same parameters as those used in the GROUP BY clause for grouping window functions.
• If event time is used, watermark must be used. The code is as follows (order_time is identified as the event time column and watermark is set to 3 seconds):

  CREATE TABLE orders (
    order_id string,
    order_channel string,
    order_time timestamp(3),
    pay_amount double,
    real_pay double,
    pay_time string,
    user_id string,
    user_name string,
    area_id string,
    watermark for order_time as order_time - INTERVAL '3' SECOND
  ) WITH (
    'connector' = 'kafka',
    'topic' = '<yourTopic>',
    'properties.bootstrap.servers' = '<yourKafka>:<port>',
    'properties.group.id' = '<yourGroupId>',
    'scan.startup.mode' = 'latest-offset',
    'format' = 'json'
  );

• If the proctime is used, you need to use the computed column. The code is as follows (proc is the processing time column):

  CREATE TABLE orders (
    order_id string,
    order_channel string,
    order_time timestamp(3),
    pay_amount double,
    real_pay double,
    pay_time string,
    user_id string,
    user_name string,
    area_id string,
    proc as proctime()
  ) WITH (
    'connector' = 'kafka',
    'topic' = '<yourTopic>',
    'properties.bootstrap.servers' = '<yourKafka>:<port>',
    'properties.group.id' = '<yourGroupId>',
    'scan.startup.mode' = 'latest-offset',
    'format' = 'json'
  );

Syntax

TUMBLE(time_attr, window_interval, period_interval, lateness_interval)

Example

TUMBLE(testtime, INTERVAL '10' SECOND, INTERVAL '10' SECOND, INTERVAL '10' SECOND)

OVER WINDOW

The difference between Over Window and Group Window is that one record is generated from one row in Over Window.

Syntax

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

SELECT agg1(attr1) OVER (
  [PARTITION BY partition_name]
  ORDER BY proctime|rowtime 
  ROWS  
 BETWEEN (UNBOUNDED|rowCOUNT) PRECEDING AND CURRENT ROW FROM TABLENAME

SELECT agg1(attr1) OVER (
  [PARTITION BY partition_name]
  ORDER BY proctime|rowtime 
  RANGE  
  BETWEEN (UNBOUNDED|timeInterval) PRECEDING AND CURRENT ROW FROM TABLENAME

Description

Precautions

• All aggregates must be defined in the same window, that is, in the same partition, sort, and range.
• Currently, only windows from PRECEDING (unbounded or bounded) to CURRENT ROW are supported. The range described by FOLLOWING is not supported.
• ORDER BY must be specified for a single time attribute.

Example

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17

// Calculate the count and total number from syntax rules enabled to now (in proctime).
insert into temp SELECT name,
    count(amount) OVER (PARTITION BY name ORDER BY proctime RANGE UNBOUNDED preceding) as cnt1,
    sum(amount) OVER (PARTITION BY name ORDER BY proctime RANGE UNBOUNDED preceding) as cnt2
    FROM Orders;
  
// Calculate the count and total number of the recent four records (in proctime).
insert into temp SELECT name,
    count(amount) OVER (PARTITION BY name ORDER BY proctime ROWS BETWEEN 4 PRECEDING AND CURRENT ROW) as cnt1,
    sum(amount) OVER (PARTITION BY name ORDER BY proctime ROWS BETWEEN 4 PRECEDING AND CURRENT ROW) as cnt2
    FROM Orders;

// Calculate the count and total number last 60s (in eventtime). Process the events based on event time, which is the timeattr field in Orders.
insert into temp SELECT name,
    count(amount) OVER (PARTITION BY name ORDER BY timeattr RANGE BETWEEN INTERVAL '60' SECOND PRECEDING AND CURRENT ROW) as cnt1,
    sum(amount) OVER (PARTITION BY name ORDER BY timeattr RANGE BETWEEN INTERVAL '60' SECOND PRECEDING AND CURRENT ROW) as cnt2
    FROM Orders;