Status

Current state:   [One of "Under Discussion", "Accepted", "Rejected"]

Discussion thread:  here (<- link to https://mail-archiveslists.apache.org/mod_mbox/flink-dev/)
JIRA: here (<- link to thread/39mwckdsdgck48tzsdfm66hhnxorjtz3

Vote thread: https://issueslists.apache.org/jira/browse/FLINK-XXXX)
thread/5fzfqc6dw6wyx2xsnh2rpsotsxhbpc26

JIRA:

Released: 1.18.0Released: <Flink Version>

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

Motivation

In FLIP-218 & FLIP-305, Flink support supports CREATE TABLE AS SELECT  statement which allows users to create new tables based on existing tables or query resultresults. It's convenient for data analysts and data scientists to manage their data. However, Flink does not currently support the REPLACE TABLE AS SELECT  statment statement which enables users to replace an existing table with new data. With REPLACE TABLE AS SELECT, they won't need to drop the table firstly, first and use CREATE TABLE AS SELECT  then. Only one single REPLACE TABLE AS SELCTSELECT  statement can meet their needs.

So, this FLIP is aimed to support REPLACE TABLE AS SELECT  statement in Flink.

Note: this FLIP is much similar to FLIP-218 & FLIP-305, you may need to read these two FLIP to get more context.

Public Interfaces

Syntax

We propose add adding the following syntax for REPLACE TABLE AS SELECT  statement:

REPLACE TABLE table_identifier
[(
 [ < physical_column_definition >, ... ]
 [ < table_constraint >, ... ])]
[ COMMENT table_comment ]
[ PARTITIONED BY ( col_name1, col_name3, ...) ]
[ WITH ( key1=val1, key2=val2, ... ) ]
AS <table subquery>

<physical_column_definition>:
  column_name column_type [ <column_constraint> ] [COMMENT column_comment]
  
<column_constraint>:
  [CONSTRAINT constraint_name] PRIMARY KEY NOT ENFORCED

<table_constraint>:
  [CONSTRAINT constraint_name] PRIMARY KEY (column_name, ...) NOT ENFORCED

Also, we would like to propose to CREATE OR REPLACE TABLE AS  to wrap CREATE TABLE AS SELECT  and REPLACE TABLE AS SELECT  which will create a table if the table to be replaced doesn't exist.

CREATE OR REPLACE TABLE table_identifier
[(
 [WITH < physical_column_definition >( key1=val1, key2=val2, ... ) ]
 [ < table_constraint >, ... ])]
[ COMMENT table_comment ]
[ PARTITIONED BY ( col_name1, col_name3, ...) ]
[ WITH ( key1=val1, key2=val2, ... ) ]
AS <table subquery>

<physical_column_definition>:
  column_name column_type [ <column_constraint> ] [COMMENT column_comment]
  
<column_constraint>:
  [CONSTRAINT constraint_name] PRIMARY KEY NOT ENFORCED

<table_constraint>:
  [CONSTRAINT constraint_name] PRIMARY KEY (column_name, ...) NOT ENFORCED

Public Interfaces Change

AS <table subquery>

Public Interfaces Change

To support atomic, we propose to add the following part to the interface SupportsStaging  proposed in FLIP-305

/**
 * Enables different staged operations to ensure atomicity in a {@link DynamicTableSink}.
 *
 * <p>By default, if this interface is not implemented, indicating that atomic operations are not
 * supported, then a non-atomic implementation is used.
 */
@PublicEvolving
public interface SupportsStaging {
    //.... emit the parts proposed in FLIP-305

    enum StagingPurpose {
        CREATE_TABLE_AS,
        CREATE_TABLE_AS_IF_NOT_EXISTS,
 The   // the following is what to add in this FLIP
        REPLACE_TABLE_AS,
        CREATE_OR_REPLACE_TABLE_AS
    }
}

Also, we propose to modify the name of the option "table.cats.atomicity-enabled" proposed in FLIP-305We propose the following public interface changes for REPLACE TABLE AS SELECT  statement:

@PublicEvolving
public interfaceclass Catalog {
  
   /** Prepare replace table. */
   Optional<TwoPhaseCommitCatalogTable> prepareReplaceTable(xxx, xxxx, xxxx, xx);

TableConfigOptions {
    @Documentation.TableOption(execMode = Documentation.ExecMode.BATCH_STREAMING)
    public static final ConfigOption<Boolean> TABLE_RTAS_CTAS_ATOMICITY_ENABLED =
            key("table.rtas-ctas.atomicity-enabled")
                    .booleanType()
                    .defaultValue(false)
                    .withDescription(
                            "Specifies if the create table/replace table/create or replace table as select operation is executed atomically. "
                                  + "By default, the operation is non-atomic. The target table is created/replaced in Client side, and it won't rollback even though the job fails or is cancelled. "
                                  + "If set this option to true and DynamicTableSink implements the SupportsStaging interface, the create table/replace table/create or replace table as select operation is expected to be executed atomically, "
                                  + "the behavior of which depends on the actual DynamicTableSink.");
}

Proposed Changes

For REPLACE TABLE AS  statement:

1:  Construct the table to be created If the column definition is specificed, it will use the defined column to  create the new table.  Otherwise, it'll retrieve the column definition from the sub-query.

2: Check the table exists or not. If the table doesn't exist, throw TableException(String.format("The table %s to be replaced doesn't exist. You may want to use CREATE TABLE AS statement or CREATE OR REPLACE TABLE AS statement.",tableIdentifier)).

3:Check the atomicity is enabled, it requires both the option table.rtas-ctas.atomicity-enabled is set to true and the corresponding table sink implements SupportsStaging.

   a: if atomic is enabled, it expects the atomicity to be guaranteed by external connector implementation. The Flink will generate an insert job according to the table subquery in REPLACE TABLE AS statment, and call method StagedTable#begain  before the insert job start, call method StagedTable#commit  after the job finish, call method StagedTable#abort  if the job fail or canceled.

   b: if not Call method  Catalog#prepareReplaceTable  to try to get the TwoPhaseCommitCatalogTable .    a: If get empty, then the atomicity can not be guaranted. Flink will do the operations for Replace Table one by one without atomicity guarantee. More exactly, it will drop the old table, create the new table, insert data into the new tables.    b: Otherwise, it expects the guarantee will be ensured to exteran connector implemter. The Flink will call method #begainTransanction before the insert job start, call method #commit after the job finish, call method #abort if the job fail or canceled.

For CREATE OR REPLACE TABLE AS  stament,  when the table exists, it'll consider it as REPLACE TABLE AS  statement. Otherwise, it'll consider it as CREATE TABLE AS  statement. 

Note: Again, the propose changes much depend on FLIP-305. For more detail, please see the proposed chagne part in FLIP-305.

Compatibility, Deprecation, and Migration Plan

...