Availability

Airbyte Cloud Airbyte OSS PyAirbyte

Support Level

Community

Latest Version

3.6.1

Microsoft SQL Server (MSSQL)

Features

Feature	Supported	Notes
Full Refresh Sync	Yes
Incremental Sync - Append	Yes
Replicate Incremental Deletes	Yes
CDC (Change Data Capture)	Yes
SSL Support	Yes
SSH Tunnel Connection	Yes
Namespaces	Yes	Enabled by default

The MSSQL source does not alter the schema present in your database. Depending on the destination connected to this source, however, the schema may be altered. See the destination's documentation for more details.

Troubleshooting

You may run into an issue where the connector provides wrong values for some data types. See discussion on unexpected behaviour for certain datatypes.

Note: Currently hierarchyid and sql_variant are not processed in CDC migration type (not supported by debezium). For more details please check this ticket

Getting Started (Airbyte Cloud)

On Airbyte Cloud, only TLS connections to your MSSQL instance are supported in source configuration. Other than that, you can proceed with the open-source instructions below.

Getting Started (Airbyte Open Source)

Requirements

1. MSSQL Server Azure SQL Database, Azure Synapse Analytics, Azure SQL Managed Instance, SQL Server 2019, SQL Server 2017, SQL Server 2016, SQL Server 2014, SQL Server 2012, PDW 2008R2 AU34.
2. Create a dedicated read-only Airbyte user with access to all tables needed for replication
3. If you want to use CDC, please see the relevant section below for further setup requirements

1. Make sure your database is accessible from the machine running Airbyte

This is dependent on your networking setup. The easiest way to verify if Airbyte is able to connect to your MSSQL instance is via the check connection tool in the UI.

2. Create a dedicated read-only user with access to the relevant tables (Recommended but optional)

This step is optional but highly recommended to allow for better permission control and auditing. Alternatively, you can use Airbyte with an existing user in your database.

Coming soon: suggestions on how to create this user.

3. Your database user should now be ready for use with Airbyte!

Change Data Capture (CDC)

We use SQL Server's change data capture feature to capture row-level INSERT, UPDATE and DELETE operations that occur on cdc-enabled tables.

Some extra setup requiring at least db_owner permissions on the database(s) you intend to sync from will be required (detailed below).

Please read the CDC docs for an overview of how Airbyte approaches CDC.

Should I use CDC for MSSQL?

• If you need a record of deletions and can accept the limitations posted below, CDC is the way to go!
• If your data set is small and/or you just want a snapshot of your table in the destination, consider using Full Refresh replication for your table instead of CDC.
• If the limitations below prevent you from using CDC and your goal is to maintain a snapshot of your table in the destination, consider using non-CDC incremental and occasionally reset the data and re-sync.
• If your table has a primary key but doesn't have a reasonable cursor field for incremental syncing (i.e. updated_at), CDC allows you to sync your table incrementally.

CDC Config

Parameter	Type	Default	Description
Data to Sync	Enum: Existing and New, New Changes Only	Existing and New	What data should be synced under the CDC. Existing and New will read existing data as a snapshot, and sync new changes through CDC. New Changes Only will skip the initial snapshot, and only sync new changes through CDC. See documentation here for details. Under the hood, this parameter sets the snapshot.mode in Debezium.
Snapshot Isolation Level	Enum: Snapshot, Read Committed	Snapshot	Mode to control which transaction isolation level is used and how long the connector locks tables that are designated for capture. If you don't know which one to choose, just use the default one. See documentation here for details. Under the hood, this parameter sets the snapshot.isolation.mode in Debezium.

CDC Limitations

• Make sure to read our CDC docs to see limitations that impact all databases using CDC replication.
• There are some critical issues regarding certain datatypes. Please find detailed info in this Github issue.
• CDC is only available for SQL Server 2016 Service Pack 1 (SP1) and later.
• db_owner (or higher) permissions are required to perform the neccessary setup for CDC.
• If you set Initial Snapshot Isolation Level to Snapshot, you must enable snapshot isolation mode on the database(s) you want to sync. This is used for retrieving an initial snapshot without locking tables.
• For SQL Server Always On read-only replica, only Snapshot initial snapshot isolation level is supported.
• On Linux, CDC is not supported on versions earlier than SQL Server 2017 CU18 (SQL Server 2019 is supported).
• Change data capture cannot be enabled on tables with a clustered columnstore index. (It can be enabled on tables with a non-clustered columnstore index).
• The SQL Server CDC feature processes changes that occur in user-created tables only. You cannot enable CDC on the SQL Server master database.
• Using variables with partition switching on databases or tables with change data capture (CDC) is not supported for the ALTER TABLE ... SWITCH TO ... PARTITION ... statement
• Our implementation has not been tested with managed instances, such as Azure SQL Database (we welcome any feedback from users who try this!)
  • If you do want to try this, CDC can only be enabled on Azure SQL databases tiers above Standard 3 (S3+). Basic, S0, S1 and S2 tiers are not supported for CDC.
• Our CDC implementation uses at least once delivery for all change records.
• Read more on CDC limitations in the Microsoft docs.

Setting up CDC for MSSQL

1. Enable CDC on database and tables

MS SQL Server provides some built-in stored procedures to enable CDC.

• To enable CDC, a SQL Server administrator with the necessary privileges (db_owner or sysadmin) must first run a query to enable CDC at the database level.

  USE {database name}
  GO
  EXEC sys.sp_cdc_enable_db
  GO

• The administrator must then enable CDC for each table that you want to capture. Here's an example:

  USE {database name}
  GO
  
  EXEC sys.sp_cdc_enable_table
  @source_schema = N'{schema name}',
  @source_name   = N'{table name}',
  @role_name     = N'{role name}',  [1]
  @filegroup_name = N'{filegroup name}', [2]
  @supports_net_changes = 0 [3]
  GO
  • [1] Specifies a role which will gain SELECT permission on the captured columns of the source table. We suggest putting a value here so you can use this role in the next step but you can also set the value of @rolename to NULL to allow only _sysadmin and db_owner to have access. Be sure that the credentials used to connect to the source in Airbyte align with this role so that Airbyte can access the cdc tables.
  • [2] Specifies the filegroup where SQL Server places the change table. We recommend creating a separate filegroup for CDC but you can leave this parameter out to use the default filegroup.
  • [3] If 0, only the support functions to query for all changes are generated. If 1, the functions that are needed to query for net changes are also generated. If supports_net_changes is set to 1, index_name must be specified, or the source table must have a defined primary key.

• (For more details on parameters, see the Microsoft doc page for this stored procedure).

• If you have many tables to enable CDC on and would like to avoid having to run this query one-by-one for every table, this script might help!

For further detail, see the Microsoft docs on enabling and disabling CDC.

2. Enable snapshot isolation

• When a sync runs for the first time using CDC, Airbyte performs an initial consistent snapshot of your database. To avoid acquiring table locks, Airbyte uses snapshot isolation, allowing simultaneous writes by other database clients. This must be enabled on the database like so:

  ALTER DATABASE {database name}
    SET ALLOW_SNAPSHOT_ISOLATION ON;

3. Create a user and grant appropriate permissions

• Rather than use sysadmin or db_owner credentials, we recommend creating a new user with the relevant CDC access for use with Airbyte. First let's create the login and user and add to the db_datareader role:

  USE {database name};
  CREATE LOGIN {user name}
    WITH PASSWORD = '{password}';
  CREATE USER {user name} FOR LOGIN {user name};
  EXEC sp_addrolemember 'db_datareader', '{user name}';

  • Add the user to the role specified earlier when enabling cdc on the table(s):

    EXEC sp_addrolemember '{role name}', '{user name}';

  • This should be enough access, but if you run into problems, try also directly granting the user SELECT access on the cdc schema:

    USE {database name};
    GRANT SELECT ON SCHEMA :: [cdc] TO {user name};

  • If feasible, granting this user 'VIEW SERVER STATE' permissions will allow Airbyte to check whether or not the SQL Server Agent is running. This is preferred as it ensures syncs will fail if the CDC tables are not being updated by the Agent in the source database.

    USE master;
    GRANT VIEW SERVER STATE TO {user name};

4. Extend the retention period of CDC data

• In SQL Server, by default, only three days of data are retained in the change tables. Unless you are running very frequent syncs, we suggest increasing this retention so that in case of a failure in sync or if the sync is paused, there is still some bandwidth to start from the last point in incremental sync.

• These settings can be changed using the stored procedure sys.sp_cdc_change_job as below:

  -- we recommend 14400 minutes (10 days) as retention period
  EXEC sp_cdc_change_job @job_type='cleanup', @retention = {minutes}

• After making this change, a restart of the cleanup job is required:

  EXEC sys.sp_cdc_stop_job @job_type = 'cleanup';

  EXEC sys.sp_cdc_start_job @job_type = 'cleanup';

5. Ensure the SQL Server Agent is running

• MSSQL uses the SQL Server Agent

  to run the jobs necessary

  for CDC. It is therefore vital that the Agent is operational in order for to CDC to work effectively. You can check

  the status of the SQL Server Agent as follows:

  EXEC xp_servicecontrol 'QueryState', N'SQLServerAGENT';

• If you see something other than 'Running.' please follow

  the Microsoft docs

  to start the service.

Connection to MSSQL via an SSH Tunnel

Airbyte has the ability to connect to a MSSQL instance via an SSH Tunnel. The reason you might want to do this because it is not possible (or against security policy) to connect to the database directly (e.g. it does not have a public IP address).

When using an SSH tunnel, you are configuring Airbyte to connect to an intermediate server (a.k.a. a bastion sever) that does have direct access to the database. Airbyte connects to the bastion and then asks the bastion to connect directly to the server.

Using this feature requires additional configuration, when creating the source. We will talk through what each piece of configuration means.

1. Configure all fields for the source as you normally would, except SSH Tunnel Method.

2. SSH Tunnel Method defaults to No Tunnel (meaning a direct connection). If you want to use an

   SSH Tunnel choose SSH Key Authentication or Password Authentication.

   1. Choose Key Authentication if you will be using an RSA private key as your secret for

      establishing the SSH Tunnel (see below for more information on generating this key).

   2. Choose Password Authentication if you will be using a password as your secret for establishing

      the SSH Tunnel.

3. SSH Tunnel Jump Server Host refers to the intermediate (bastion) server that Airbyte will connect to. This should

   be a hostname or an IP Address.

4. SSH Connection Port is the port on the bastion server with which to make the SSH connection. The default port for

   SSH connections is 22, so unless you have explicitly changed something, go with the default.

5. SSH Login Username is the username that Airbyte should use when connection to the bastion server. This is NOT the

   MSSQL username.

6. If you are using Password Authentication, then SSH Login Username should be set to the

   password of the User from the previous step. If you are using SSH Key Authentication leave this

   blank. Again, this is not the MSSQL password, but the password for the OS-user that Airbyte is

   using to perform commands on the bastion.

7. If you are using SSH Key Authentication, then SSH Private Key should be set to the RSA

   private Key that you are using to create the SSH connection. This should be the full contents of

   the key file starting with -----BEGIN RSA PRIVATE KEY----- and ending

   with -----END RSA PRIVATE KEY-----.

Generating an SSH Key Pair

The connector expects an RSA key in PEM format. To generate this key:

ssh-keygen -t rsa -m PEM -f myuser_rsa

This produces the private key in pem format, and the public key remains in the standard format used by the authorized_keys file on your bastion host. The public key should be added to your bastion host to whichever user you want to use with Airbyte. The private key is provided via copy-and-paste to the Airbyte connector configuration screen, so it may log in to the bastion.

Data type mapping

MSSQL data types are mapped to the following data types when synchronizing data. You can check the test values examples here. If you can't find the data type you are looking for or have any problems feel free to add a new test!

MSSQL Type	Resulting Type	Notes
bigint	number
binary	string
bit	boolean
char	string
date	date
datetime	timestamp
datetime2	timestamp
datetimeoffset	timestamp with timezone
decimal	number
int	number
float	number
geography	string
geometry	string
money	number
numeric	number
ntext	string
nvarchar	string
nvarchar(max)	string
real	number
smalldatetime	timestamp
smallint	number
smallmoney	number
sql_variant	string
uniqueidentifier	string
text	string
time	time
tinyint	number
varbinary	string
varchar	string
varchar(max) COLLATE Latin1_General_100_CI_AI_SC_UTF8	string
xml	string

If you do not see a type in this list, assume that it is coerced into a string. We are happy to take feedback on preferred mappings.

Upgrading from 0.4.17 and older versions to 0.4.18 and newer versions

There is a backwards incompatible spec change between Microsoft SQL Source connector versions 0.4.17 and 0.4.18. As part of that spec change replication_method configuration parameter was changed to object from string.

In Microsoft SQL source connector versions 0.4.17 and older, replication_method configuration parameter was saved in the configuration database as follows:

"replication_method": "STANDARD"

Starting with version 0.4.18, replication_method configuration parameter is saved as follows:

"replication_method": {
    "method": "STANDARD"
}

After upgrading Microsoft SQL Source connector from 0.4.17 or older version to 0.4.18 or newer version you need to fix source configurations in the actor table in Airbyte database. To do so, you need to run two SQL queries. Follow the instructions in Airbyte documentation to run SQL queries on Airbyte database.

If you have connections with Microsoft SQL Source using Standard replication method, run this SQL:

update public.actor set configuration =jsonb_set(configuration, '{replication_method}', '{"method": "STANDARD"}', true)
WHERE actor_definition_id ='b5ea17b1-f170-46dc-bc31-cc744ca984c1' AND (configuration->>'replication_method' = 'STANDARD');

If you have connections with Microsoft SQL Source using Logicai Replication (CDC) method, run this SQL:

update public.actor set configuration =jsonb_set(configuration, '{replication_method}', '{"method": "CDC"}', true)
WHERE actor_definition_id ='b5ea17b1-f170-46dc-bc31-cc744ca984c1' AND (configuration->>'replication_method' = 'CDC');

Changelog

Version	Date	Pull Request	Subject
3.7.5	2024-02-29	35739	Allow configuring the queue size used for cdc events.
3.7.4	2024-02-26	35566	Add config to throw an error on invalid CDC position.
3.7.3	2024-02-23	35596	Fix a logger issue
3.7.2	2024-02-21	35368	Change query syntax to make it compatible with Azure SQL Managed Instance.
3.7.1	2024-02-20	35405	Change query syntax to make it compatible with Azure Synapse.
3.7.0	2024-01-30	33311	Source mssql with checkpointing initial sync.
3.6.1	2024-01-26	34573	Adopt CDK v0.16.0.
3.6.0	2024-01-10	33700	Remove CDC config options for data_to_sync and snapshot isolation.
3.5.1	2024-01-05	33510	Test-only changes.
3.5.0	2023-12-19	33071	Fix SSL configuration parameters
3.4.1	2024-01-02	33755	Encode binary to base64 format
3.4.0	2023-12-19	33481	Remove LEGACY state flag
3.3.2	2023-12-14	33505	Using the released CDK.
3.3.1	2023-12-12	33225	extracting MsSql specific files out of the CDK.
3.3.0	2023-12-12	33018	Migrate to Per-stream/Global states and away from Legacy states
3.2.1	2023-12-11	33330	Parse DatetimeOffset fields with the correct format when used as cursor
3.2.0	2023-12-07	33225	CDC : Enable compression of schema history blob in state.
3.1.0	2023-11-28	32882	Enforce SSL on Airbyte Cloud.
3.0.2	2023-11-27	32573	Format Datetime and Datetime2 datatypes to 6-digit microsecond precision
3.0.1	2023-11-22	32656	Adopt java CDK version 0.5.0.
3.0.0	2023-11-07	31531	Remapped date, smalldatetime, datetime2, time, and datetimeoffset datatype to their correct Airbyte types
2.0.4	2023-11-06	#32193	Adopt java CDK version 0.4.1.
2.0.3	2023-10-31	32024	Upgrade to Debezium version 2.4.0.
2.0.2	2023-10-30	31960	Adopt java CDK version 0.2.0.
2.0.1	2023-08-24	29821	Set replication_method display_type to radio, update titles and descriptions, and make CDC the default choice
2.0.0	2023-08-22	29493	Set a default cursor for Cdc mode
1.1.1	2023-07-24	28545	Support Read Committed snapshot isolation level
1.1.0	2023-06-26	27737	License Update: Elv2
1.0.19	2023-06-20	27212	Fix silent exception swallowing in StreamingJdbcDatabase
1.0.18	2023-06-14	27335	Remove noisy debug logs
1.0.17	2023-05-25	26473	CDC : Limit queue size
1.0.16	2023-05-01	25740	Disable index logging
1.0.15	2023-04-26	25401	CDC : Upgrade Debezium to version 2.2.0
1.0.14	2023-04-19	25345	Logging : Log database indexes per stream
1.0.13	2023-04-19	24582	CDC : refactor for performance improvement
1.0.12	2023-04-17	25220	Logging changes : Log additional metadata & clean up noisy logs
1.0.11	2023-04-11	24656	CDC minor refactor
1.0.10	2023-04-06	24820	Fix data loss bug during an initial failed non-CDC incremental sync
1.0.9	2023-04-04	24833	Fix Debezium retry policy configuration
1.0.8	2023-03-28	24166	Fix InterruptedException bug during Debezium shutdown
1.0.7	2023-03-27	24529	Preparing the connector for CDC checkpointing
1.0.6	2023-03-22	20760	Removed redundant date-time datatypes formatting
1.0.5	2023-03-21	24207	Fix incorrect schema change warning in CDC mode
1.0.4	2023-03-21	24147	Fix error with CDC checkpointing
1.0.3	2023-03-15	24082	Fixed NPE during cursor values validation
1.0.2	2023-03-14	23908	Log warning on null cursor values
1.0.1	2023-03-10	23939	For network isolation, source connector accepts a list of hosts it is allowed to connect
1.0.0	2023-03-06	23112	Upgrade Debezium version to 2.1.2
0.4.29	2023-02-24	16798	Add event_serial_no to cdc metadata
0.4.28	2023-01-18	21348	Fix error introduced in 18959 in which option initial_waiting_seconds was removed
0.4.27	2022-12-14	20436	Consolidate date/time values mapping for JDBC sources
0.4.26	2022-12-12	18959	CDC : Don't timeout if snapshot is not complete.
0.4.25	2022-11-04	18732	Upgrade debezium version to 1.9.6
0.4.24	2022-10-25	18383	Better SSH error handling + messages
0.4.23	2022-10-21	18263	Fixes bug introduced in 15833 and adds better error messaging for SSH tunnel in Destinations
0.4.22	2022-10-19	18087	Better error messaging for configuration errors (SSH configs, choosing an invalid cursor)
0.4.21	2022-10-17	18041	Fixes bug introduced 2022-09-12 with SshTunnel, handles iterator exception properly
	2022-10-13	15535	Update incremental query to avoid data missing when new data is inserted at the same time as a sync starts under non-CDC incremental mode
0.4.20	2022-09-14	15668	Wrap logs in AirbyteLogMessage
0.4.19	2022-09-05	16002	Added ability to specify schemas for discovery during setting connector up
0.4.18	2022-09-03	14910	Standardize spec for CDC replication. Replace the replication_method enum with a config object with a method enum field.
0.4.17	2022-09-01	16261	Emit state messages more frequently
0.4.16	2022-08-18	14356	DB Sources: only show a table can sync incrementally if at least one column can be used as a cursor field
0.4.15	2022-08-11	15538	Allow additional properties in db stream state
0.4.14	2022-08-10	15430	fixed a bug on handling special character on database name
0.4.13	2022-08-04	15268	Added [] enclosing to escape special character in the database name
0.4.12	2022-08-02	14801	Fix multiple log bindings
0.4.11	2022-07-22	14714	Clarified error message when invalid cursor column selected
0.4.10	2022-07-14	14574	Removed additionalProperties:false from JDBC source connectors
0.4.9	2022-07-05	14379	Aligned Normal and CDC migration + added some fixes for datatypes processing
0.4.8	2022-06-24	14121	Omit using 'USE' keyword on Azure SQL with CDC
0.4.5	2022-06-23	14077	Use the new state management
0.4.3	2022-06-17	13887	Increase version to include changes from 13854
0.4.2	2022-06-06	13435	Adjust JDBC fetch size based on max memory and max row size
0.4.1	2022-05-25	13419	Correct enum for Standard method.
0.4.0	2022-05-25	12759 13168	For CDC, Add option to ignore existing data and only sync new changes from the database.
0.3.22	2022-04-29	12480	Query tables with adaptive fetch size to optimize JDBC memory consumption
0.3.21	2022-04-11	11729	Bump mina-sshd from 2.7.0 to 2.8.0
0.3.19	2022-03-31	11495	Adds Support to Chinese MSSQL Server Agent
0.3.18	2022-03-29	11010	Adds JDBC Params
0.3.17	2022-02-21	10242	Fixed cursor for old connectors that use non-microsecond format. Now connectors work with both formats
0.3.16	2022-02-18	10242	Updated timestamp transformation with microseconds
0.3.15	2022-02-14	10256	Add -XX:+ExitOnOutOfMemoryError JVM option
0.3.14	2022-01-24	9554	Allow handling of java sql date in CDC
0.3.13	2022-01-07	9094	Added support for missed data types
0.3.12	2021-12-30	9206	Update connector fields title/description
0.3.11	2021-12-24	8958	Add support for JdbcType.ARRAY
0.3.10	2021-12-01	8371	Fixed incorrect handling "\n" in ssh key
0.3.9	2021-11-09	7748	Improve support for binary and varbinary data types
0.3.8	2021-10-26	7386	Fixed data type (smalldatetime, smallmoney) conversion from mssql source
0.3.7	2021-09-30	6585	Improved SSH Tunnel key generation steps
0.3.6	2021-09-17	6318	Added option to connect to DB via SSH
0.3.4	2021-08-13	4699	Added json config validator
0.3.3	2021-07-05	4689	Add CDC support
0.3.2	2021-06-09	3179	Add AIRBYTE_ENTRYPOINT for Kubernetes support
0.3.1	2021-06-08	3893	Enable SSL connection
0.3.0	2021-04-21	2990	Support namespaces
0.2.3	2021-03-28	2600	Add NCHAR and NVCHAR support to DB and cursor type casting
0.2.2	2021-03-26	2460	Destination supports destination sync mode
0.2.1	2021-03-18	2488	Sources support primary keys
0.2.0	2021-03-09	2238	Protocol allows future/unknown properties
0.1.11	2021-02-02	1887	Migrate AbstractJdbcSource to use iterators
0.1.10	2021-01-25	1746	Fix NPE in State Decorator
0.1.9	2021-01-19	1724	Fix JdbcSource handling of tables with same names in different schemas
0.1.9	2021-01-14	1655	Fix JdbcSource OOM
0.1.8	2021-01-13	1588	Handle invalid numeric values in JDBC source
0.1.6	2020-12-09	1172	Support incremental sync
0.1.5	2020-11-30	1038	Change JDBC sources to discover more than standard schemas
0.1.4	2020-11-30	1046	Add connectors using an index YAML file