SELECT Statement

Confluent Cloud for Apache Flink®️ enables querying the content of your tables by using familiar SELECT syntax.

Syntax

SELECT [DISTINCT] select_list FROM table_expression [ WHERE boolean_expression ]

Description

The SELECT statement in Flink does what the SQL standard says it must do. You needn’t look further than standard SQL itself to understand the behavior.

For example, UNION without ALL means that duplicate rows must be removed.

Flink maintains the relation, called a dynamic table, specified by the SQL query. Its behavior is always the same as if you ran the SQL query again, over the current snapshot of the data, each time a new row arrives for any table in the relation.

This formalism is what enables you to reason about exactly what Flink will do just by understanding what any SQL system, like MySQL, Snowflake, or Oracle, would do.

Another way to understand what Flink SQL does is to consider the following statement:

SELECT * FROM clicks ORDER BY clickTime LIMIT 10;

This statement doesn’t only look at 10 rows, sort them, and terminate. It maintains this relation, and as new orders arrive, the relation changes, always showing the top 10 most recent orders. This is exactly as if you re-ran the query each time a new row was written to the clicks table. You’ll get the same result.

Select list

The select_list specification * means the query resolves all columns. But in production, using * is not recommended, because it makes queries less robust to catalog changes. Instead, use a select_list to specify a subset of available columns or make calculations using the columns. For example, if an orders table has columns named order_id, price, and tax you could write the following query:

SELECT order_id, price + tax FROM orders

Table expression

The table_expression can be any source of data, including a table, view, or VALUES clause, the joined results of multiple existing tables, or a subquery.

Assuming that an orders table is available in the catalog, the following would read all rows from .

SELECT * FROM orders;

VALUES clause

Queries can consume from inline data by using the VALUES clause. Each tuple corresponds to one row. You can provide an alias to assign a name to each column.

 SELECT order_id, price
   FROM (VALUES (1, 2.0), (2, 3.1))
   AS t (order_id, price);

Your output should resemble:

order_id price
1        2.0
2        3.1

WHERE clause

Filter rows by using the WHERE clause.

 SELECT price + tax
   FROM orders
   WHERE id = 10;

Functions

You can invoke built-in scalar functions on the columns of a single row.

SELECT PRETTY_PRINT(order_id) FROM orders;

DISTINCT

If SELECT DISTINCT is specified, all duplicate rows are removed from the result set, which means that one row is kept from each group of duplicates.

For streaming queries, the required state for computing the query result might grow infinitely. State size depends on the number of distinct rows.

SELECT DISTINCT id FROM orders;

Examples

In the Flink SQL shell or in a Cloud Console workspace, run the following commands to see examples of the SELECT statement.

  1. Create a table for web page click events.

    -- Create a table for web page click events.
CREATE TABLE clicks (
  ip_address VARCHAR,
  url VARCHAR,
  click_ts_raw BIGINT
);

  2. Populate the table with mock clickstream data.

    -- Populate the table with mock clickstream data.
INSERT INTO clicks
VALUES( '10.0.0.1',  'https://acme.com/index.html',     1692812175),
      ( '10.0.0.12', 'https://apache.org/index.html',   1692826575),
      ( '10.0.0.13', 'https://confluent.io/index.html', 1692826575),
      ( '10.0.0.1',  'https://acme.com/index.html',     1692812175),
      ( '10.0.0.12', 'https://apache.org/index.html',   1692819375),
      ( '10.0.0.13', 'https://confluent.io/index.html', 1692826575);

    Press ENTER to return to the SQL shell. Because INSERT INTO VALUES is a point-in-time statement, it exits after it completes inserting records.

  3. View all rows in the clicks table by using a SELECT statement.

    SELECT * FROM clicks;

    Your output should resemble:

    ip_address url                             click_ts_raw
10.0.0.1   https://acme.com/index.html     1692812175
10.0.0.12  https://apache.org/index.html   1692826575
10.0.0.13  https://confluent.io/index.html 1692826575
10.0.0.1   https://acme.com/index.html     1692812175
10.0.0.12  https://apache.org/index.html   1692819375
10.0.0.13  https://confluent.io/index.html 1692826575

  4. View only unique rows in the clicks table by using a SELECT DISTINCT statement.

    SELECT DISTINCT * FROM clicks;

    Your output should resemble:

    ip_address url                             click_ts_raw
10.0.0.1   https://acme.com/index.html     1692812175
10.0.0.12  https://apache.org/index.html   1692826575
10.0.0.13  https://confluent.io/index.html 1692826575
10.0.0.12  https://apache.org/index.html   1692819375

  5. View only records that have the ip_address of 10.0.0.1 by using a SELECT WHERE statement.

    SELECT * FROM clicks WHERE ip_address='10.0.0.1';

    Your output should resemble:

    ip_address url                         click_ts_raw
10.0.0.1   https://acme.com/index.html 1692812175
10.0.0.1   https://acme.com/index.html 1692812175

SELECT examples

Most minimal statement

Syntax
SELECT 1;
Properties
  • Statement is bounded

Check local time zone is configured correctly

Syntax
SELECT NOW();
Properties
  • Statement is bounded
  • NOW() returns a TIMSTAMP_LTZ(3), so if the client is configured correctly, it should show a timestamp in your local time zone.

Combine multiple tables into one

Syntax
CREATE TABLE t_union_1 (i INT);
CREATE TABLE t_union_2 (i INT);
TABLE t_union_1 UNION ALL TABLE t_union_2;

-- alternate syntax
SELECT * FROM t_union_1
UNION ALL
SELECT * FROM t_union_2;

Related content

Note

This website includes content developed at the Apache Software Foundation under the terms of the Apache License v2.