Data type mapping

TwinCommit maps each source column type to the closest target type when twincommit ddl creates tables. Values are transferred losslessly unless a note says otherwise.

Db2 → targets

Db2	PostgreSQL	SQL Server	Databricks (Delta)	Snowflake
SMALLINT, INTEGER, BIGINT	smallint, integer, bigint	smallint, int, bigint	SMALLINT, INT, BIGINT	NUMBER(5/10/19,0)
DECIMAL(p,s)	numeric(p,s)	decimal(p,s)	DECIMAL(p,s)	NUMBER(p,s)
REAL, DOUBLE	real, double precision	real, float(53)	FLOAT, DOUBLE	FLOAT
CHAR(n), VARCHAR(n)	char(n), varchar(n)	nchar(n), nvarchar(n); nvarchar(max) above 4000	STRING	VARCHAR(n)
CHAR/VARCHAR FOR BIT DATA, BINARY, VARBINARY	bytea	binary(n), varbinary(n); varbinary(max) above 8000	BINARY	BINARY
DATE	date	date	DATE	DATE
TIME	time(0)	time(0)	STRING (HH:MM:SS)	TIME(0)
TIMESTAMP(p)	timestamp(p), max 6 digits (rounded)	datetime2(p), max 7 digits (truncated)	TIMESTAMP_NTZ (µs)	TIMESTAMP_NTZ(p)
BOOLEAN	boolean	bit	BOOLEAN	BOOLEAN

LOB, XML, DECFLOAT and GRAPHIC columns must be listed under exclude_columns.

Oracle → targets

Oracle	PostgreSQL	SQL Server	Databricks (Delta)	Snowflake
NUMBER(p,s)	numeric(p,s)	decimal(p,s)	DECIMAL(p,s)	NUMBER(p,s)
NUMBER (no precision)	numeric	decimal(38,10)	DECIMAL(38,10)	NUMBER(38,10)
BINARY_FLOAT, BINARY_DOUBLE	real, double precision	real, float(53)	FLOAT, DOUBLE	FLOAT
VARCHAR2(n), NVARCHAR2(n), CHAR(n)	varchar(n), char(n)	nvarchar(n), nchar(n)	STRING	VARCHAR(n)
DATE	timestamp(0)	datetime2(0)	TIMESTAMP_NTZ	TIMESTAMP_NTZ(0)
TIMESTAMP(p)	timestamp(p)	datetime2(p)	TIMESTAMP_NTZ	TIMESTAMP_NTZ(p)
TIMESTAMP WITH TIME ZONE	timestamptz	datetimeoffset	TIMESTAMP	TIMESTAMP_TZ
RAW(n)	bytea	varbinary(n)	BINARY	BINARY
CLOB, NCLOB (≤ 16 MiB)	text	nvarchar(max)	STRING	VARCHAR
BLOB (≤ 16 MiB)	bytea	varbinary(max)	BINARY	BINARY

SQL Server → targets

SQL Server	PostgreSQL	Databricks (Delta)	Snowflake
tinyint, smallint, int, bigint	smallint, smallint, integer, bigint	TINYINT, SMALLINT, INT, BIGINT	NUMBER
decimal(p,s), money	numeric(p,s), numeric(19,4)	DECIMAL	NUMBER
nvarchar(n), varchar(n)	varchar(n)	STRING	VARCHAR(n)
nvarchar(max), varchar(max)	text	STRING	VARCHAR
datetime2(p), datetime	timestamp(p), timestamp(3)	TIMESTAMP_NTZ	TIMESTAMP_NTZ
datetimeoffset	timestamptz	TIMESTAMP	TIMESTAMP_TZ
uniqueidentifier	uuid	STRING	VARCHAR(36)
bit	boolean	BOOLEAN	BOOLEAN
varbinary(n)	bytea	BINARY	BINARY

MySQL and MariaDB → targets

MySQL / MariaDB	PostgreSQL	SQL Server	Databricks (Delta)	Snowflake / BigQuery
TINYINT, SMALLINT, INT, BIGINT (signed)	smallint, smallint, integer, bigint	tinyint, smallint, int, bigint	TINYINT … BIGINT	NUMBER / INT64
BIGINT UNSIGNED	numeric(20,0)	decimal(20,0)	DECIMAL(20,0)	NUMBER(20,0) / NUMERIC
DECIMAL(p,s)	numeric(p,s)	decimal(p,s)	DECIMAL(p,s)	NUMBER(p,s) / NUMERIC or BIGNUMERIC
FLOAT, DOUBLE	real, double precision	real, float(53)	FLOAT, DOUBLE	FLOAT / FLOAT64
CHAR(n), VARCHAR(n)	char(n), varchar(n)	nchar(n), nvarchar(n)	STRING	VARCHAR(n) / STRING
TEXT, MEDIUMTEXT, LONGTEXT	text	nvarchar(max)	STRING	VARCHAR / STRING
BLOB family, BINARY, VARBINARY	bytea	varbinary(max)	BINARY	BINARY / BYTES
DATETIME(p), TIMESTAMP(p)	timestamp(p), timestamptz(p)	datetime2(p), datetimeoffset	TIMESTAMP_NTZ, TIMESTAMP	TIMESTAMP_NTZ, TIMESTAMP_TZ / DATETIME, TIMESTAMP
DATE, TIME	date, time	date, time	DATE, STRING	DATE, TIME
JSON	jsonb	nvarchar(max)	STRING	VARIANT / JSON
ENUM, SET	text	nvarchar(255)	STRING	VARCHAR / STRING
BIT(1) / BOOLEAN	boolean	bit	BOOLEAN	BOOLEAN / BOOL

Zero dates (0000-00-00) are written as NULL. Set zero_date: epoch on the table to use 1970-01-01 instead.

SAP HANA → targets

SAP HANA	PostgreSQL	SQL Server	Databricks (Delta)	Snowflake
TINYINT, SMALLINT, INTEGER, BIGINT	smallint, smallint, integer, bigint	tinyint, smallint, int, bigint	TINYINT … BIGINT	NUMBER
DECIMAL(p,s), SMALLDECIMAL	numeric(p,s), numeric	decimal(p,s), decimal(38,10)	DECIMAL	NUMBER
NVARCHAR(n), VARCHAR(n), ALPHANUM(n)	varchar(n)	nvarchar(n)	STRING	VARCHAR(n)
NCLOB, CLOB (≤ 16 MiB)	text	nvarchar(max)	STRING	VARCHAR
VARBINARY(n), BLOB (≤ 16 MiB)	bytea	varbinary	BINARY	BINARY
DATE, TIME, SECONDDATE, TIMESTAMP	date, time, timestamp(0), timestamp(6)	date, time, datetime2(0), datetime2(7)	DATE, STRING, TIMESTAMP_NTZ	DATE, TIME, TIMESTAMP_NTZ
BOOLEAN	boolean	bit	BOOLEAN	BOOLEAN

ABAP DATS and TIMS fields stored as NVARCHAR(8) / NVARCHAR(6) are replicated as strings. Add cast: {ERDAT: date} on the table to convert them.

MongoDB → relational targets

With flatten: true, top-level fields map to columns. Without it, the whole document is stored in a single JSON column next to the _id key.

BSON	PostgreSQL	SQL Server	Snowflake / BigQuery	Databricks
ObjectId	text (24 hex)	char(24)	VARCHAR(24) / STRING	STRING
int32, int64	integer, bigint	int, bigint	NUMBER / INT64	INT, BIGINT
double, decimal128	double precision, numeric	float(53), decimal(38,10)	FLOAT, NUMBER / FLOAT64, BIGNUMERIC	DOUBLE, DECIMAL(38,10)
string	text	nvarchar(max)	VARCHAR / STRING	STRING
date	timestamptz(3)	datetimeoffset(3)	TIMESTAMP_TZ / TIMESTAMP	TIMESTAMP
bool	boolean	bit	BOOLEAN / BOOL	BOOLEAN
object, array	jsonb	nvarchar(max) (JSON)	VARIANT / JSON	STRING (JSON)
binData	bytea	varbinary(max)	BINARY / BYTES	BINARY

Relational → MongoDB, Cassandra, Redis and Elasticsearch

• MongoDB: numbers map to int32, int64 or decimal128; timestamps map to date; binary maps to binData; DECIMAL keeps its precision as decimal128.
• Cassandra: the column types follow CQL (int, bigint, decimal, text, timestamp, blob, boolean). The source primary key becomes the partition key unless partition_key is set on the table.
• Redis: every value is stored as a string in the hash, with timestamps in ISO 8601 and decimals as exact strings, or as native JSON types when json: true.
• Elasticsearch: twincommit ddl generates an index template with keyword, long, scaled_float/double, date and binary fields.

Note

Kafka targets carry values in the chosen format: logical types in Avro and Protobuf, and strings in ISO 8601 or decimal-string form in JSON.