This topic describes how to create a data view in Realtime Compute for Apache Flink to simplify the development process.

Background information

If your business logic is complex, you must write nested statements in a DML statement, which makes it difficult to locate a problem. To simplify the development process, you can define a data view and write nested statements to the data view.
Note A data view displays a logical table that describes the computing logic. It does not physically store data.

Syntax

CREATE VIEW viewName[ (columnName[ , columnName]* ) ] AS queryStatement;
  • viewName: the name of the view.
  • columnName: the name of the field.
  • queryStatement: the alias of the nested statement.

Example 1

CREATE VIEW LargeOrders (r, t, c, u) AS
SELECT
    rowtime,
    productId,
    c,
    units
FROM
    orders;

INSERT INTO
    rds_output
SELECT
    r,
    t,
    c,
    u
FROM
    LargeOrders;

Example 2

  • Test data
    a (VARCHAR) b (BIGINT) c (TIMESTAMP)
    test1 1 1506823820000
    test2 1 1506823850000
    test1 1 1506823810000
    test2 1 1506823840000
    test2 1 1506823870000
    test1 1 1506823830000
    test2 1 1506823860000
  • Test statements 
    CREATE TABLE datahub_stream (
   a VARCHAR,
   b BIGINT,
   c TIMESTAMP,
   d AS PROCTIME()
) WITH (
  TYPE='datahub',
  ...
);

CREATE TABLE rds_output (
   a VARCHAR,
   b TIMESTAMP, 
   cnt BIGINT,
   PRIMARY KEY(a)
)WITH(
  TYPE = 'rds',
  ...
);

CREATE VIEW rds_view AS
SELECT a, 
   CAST(
      HOP_START(d, INTERVAL '5' SECOND, INTERVAL '30' SECOND) AS TIMESTAMP
   ) AS cc, 
   SUM(b) AS cnt
FROM 
   datahub_stream 
GROUP BY
    HOP(d, INTERVAL '5' SECOND, INTERVAL '30' SECOND),a;

INSERT INTO 
   rds_output
SELECT
   a,
   cc,
   cnt
FROM 
   rds_view
WHERE 
   cnt=4;
  • Test results
    a(VARCHAR) b (TIMESTAMP) cnt (BIGINT)
    test2 2017-11-06 16:54:10 4