Debezium connector for SQL Server

Overview

The Debezium SQL Server connector is based on the change data capture feature that is available in SQL Server 2016 Service Pack 1 (SP1) and later Standard edition or Enterprise edition. The SQL Server capture process monitors designated databases and tables, and stores the changes into specifically created change tables that have stored procedure facades.

To enable the Debezium SQL Server connector to capture change event records for database operations, you must first enable change data capture on the SQL Server database. CDC must be enabled on both the database and on each table that you want to capture. After you set up CDC on the source database, the connector can capture row-level INSERT, UPDATE, and DELETE operations that occur in the database. The connector writes event records for each source table to a Kafka topic especially dedicated to that table. One topic exists for each captured table. Client applications read the Kafka topics for the database tables that they follow, and can respond to the row-level events they consume from those topics.

The first time that the connector connects to a SQL Server database or cluster, it takes a consistent snapshot of the schemas for all tables for which it is configured to capture changes, and streams this state to Kafka. After the snapshot is complete, the connector continuously captures subsequent row-level changes that occur. By first establishing a consistent view of all of the data, the connector can continue reading without having lost any of the changes that were made while the snapshot was taking place.

The Debezium SQL Server connector is tolerant of failures. As the connector reads changes and produces events, it periodically records the position of events in the database log (LSN / Log Sequence Number). If the connector stops for any reason (including communication failures, network problems, or crashes), after a restart the connector resumes reading the SQL Server CDC tables from the last point that it read.

Fault tolerance also applies to snapshots. That is, if the connector stops during a snapshot, the connector begins a new snapshot when it restarts.

How the SQL Server connector works

To optimally configure and run a Debezium SQL Server connector, it is helpful to understand how the connector performs snapshots, streams change events, determines Kafka topic names, and uses metadata.

Snapshots

SQL Server CDC is not designed to store a complete history of database changes. For the Debezium SQL Server connector to establish a baseline for the current state of the database, it uses a process called snapshotting.

You can configure how the connector creates snapshots. By default, the connector’s snapshot mode is set to initial. Based on this initial snapshot mode, the first time that the connector starts, it performs an initial consistent snapshot of the database. This initial snapshot captures the structure and data for any tables that match the criteria defined by the include and exclude properties that are configured for the connector (for example, table.include.list, column.include.list, table.exclude.list, and so forth).

When the connector creates a snapshot, it completes the following tasks:

1. Determines the tables to be captured.

2. Obtains a lock on the SQL Server tables for which CDC is enabled to prevent structural changes from occurring during creation of the snapshot. The level of the lock is determined by snapshot.isolation.mode configuration option.

3. Reads the maximum log sequence number (LSN) position in the server’s transaction log.

4. Captures the structure of all relevant tables.

5. Releases the locks obtained in Step 2, if necessary. In most cases, locks are held for only a short period of time.

6. Scans the SQL Server source tables and schemas to be captured based on the LSN position that was read in Step 3, generates a READ event for each row in the table, and writes the events to the Kafka topic for the table.

7. Records the successful completion of the snapshot in the connector offsets.

The resulting initial snapshot captures the current state of each row in the tables that are enabled for CDC. From this baseline state, the connector captures subsequent changes as they occur.

Ad hoc snapshots

By default, a connector runs an initial snapshot operation only after it starts for the first time. Following this initial snapshot, under normal circumstances, the connector does not repeat the snapshot process. Any future change event data that the connector captures comes in through the streaming process only.

However, in some situations the data that the connector obtained during the initial snapshot might become stale, lost, or incomplete. To provide a mechanism for recapturing table data, Debezium includes an option to perform ad hoc snapshots. The following changes in a database might be cause for performing an ad hoc snapshot:

• The connector configuration is modified to capture a different set of tables.

• Kafka topics are deleted and must be rebuilt.

• Data corruption occurs due to a configuration error or some other problem.

You can re-run a snapshot for a table for which you previously captured a snapshot by initiating a so-called ad-hoc snapshot. Ad hoc snapshots require the use of signaling tables. You initiate an ad hoc snapshot by sending a signal request to the Debezium signaling table.

When you initiate an ad hoc snapshot of an existing table, the connector appends content to the topic that already exists for the table. If a previously existing topic was removed, Debezium can create a topic automatically if automatic topic creation is enabled.

Ad hoc snapshot signals specify the tables to include in the snapshot. The snapshot can capture the entire contents of the database, or capture only a subset of the tables in the database. Also, the snapshot can capture a subset of the contents of the table(s) in the database.

You specify the tables to capture by sending an execute-snapshot message to the signaling table. Set the type of the execute-snapshot signal to incremental, and provide the names of the tables to include in the snapshot, as described in the following table:

Triggering an ad hoc snapshot

You initiate an ad hoc snapshot by adding an entry with the execute-snapshot signal type to the signaling table. After the connector processes the message, it begins the snapshot operation. The snapshot process reads the first and last primary key values and uses those values as the start and end point for each table. Based on the number of entries in the table, and the configured chunk size, Debezium divides the table into chunks, and proceeds to snapshot each chunk, in succession, one at a time.

Currently, the execute-snapshot action type triggers incremental snapshots only. For more information, see Incremental snapshots.

Incremental snapshots

To provide flexibility in managing snapshots, Debezium includes a supplementary snapshot mechanism, known as incremental snapshotting. Incremental snapshots rely on the Debezium mechanism for sending signals to a Debezium connector. Incremental snapshots are based on the DDD-3 design document.

In an incremental snapshot, instead of capturing the full state of a database all at once, as in an initial snapshot, Debezium captures each table in phases, in a series of configurable chunks. You can specify the tables that you want the snapshot to capture and the size of each chunk. The chunk size determines the number of rows that the snapshot collects during each fetch operation on the database. The default chunk size for incremental snapshots is 1 KB.

As an incremental snapshot proceeds, Debezium uses watermarks to track its progress, maintaining a record of each table row that it captures. This phased approach to capturing data provides the following advantages over the standard initial snapshot process:

• You can run incremental snapshots in parallel with streamed data capture, instead of postponing streaming until the snapshot completes. The connector continues to capture near real-time events from the change log throughout the snapshot process, and neither operation blocks the other.

• If the progress of an incremental snapshot is interrupted, you can resume it without losing any data. After the process resumes, the snapshot begins at the point where it stopped, rather than recapturing the table from the beginning.

• You can run an incremental snapshot on demand at any time, and repeat the process as needed to adapt to database updates. For example, you might re-run a snapshot after you modify the connector configuration to add a table to its table.include.list property.

Incremental snapshot process

When you run an incremental snapshot, Debezium sorts each table by primary key and then splits the table into chunks based on the configured chunk size. Working chunk by chunk, it then captures each table row in a chunk. For each row that it captures, the snapshot emits a READ event. That event represents the value of the row when the snapshot for the chunk began.

As a snapshot proceeds, it’s likely that other processes continue to access the database, potentially modifying table records. To reflect such changes, INSERT, UPDATE, or DELETE operations are committed to the transaction log as per usual. Similarly, the ongoing Debezium streaming process continues to detect these change events and emits corresponding change event records to Kafka.

How Debezium resolves collisions among records with the same primary key

In some cases, the UPDATE or DELETE events that the streaming process emits are received out of sequence. That is, the streaming process might emit an event that modifies a table row before the snapshot captures the chunk that contains the READ event for that row. When the snapshot eventually emits the corresponding READ event for the row, its value is already superseded. To ensure that incremental snapshot events that arrive out of sequence are processed in the correct logical order, Debezium employs a buffering scheme for resolving collisions. Only after collisions between the snapshot events and the streamed events are resolved does Debezium emit an event record to Kafka.

Snapshot window

To assist in resolving collisions between late-arriving READ events and streamed events that modify the same table row, Debezium employs a so-called snapshot window. The snapshot windows demarcates the interval during which an incremental snapshot captures data for a specified table chunk. Before the snapshot window for a chunk opens, Debezium follows its usual behavior and emits events from the transaction log directly downstream to the target Kafka topic. But from the moment that the snapshot for a particular chunk opens, until it closes, Debezium performs a de-duplication step to resolve collisions between events that have the same primary key..

For each data collection, the Debezium emits two types of events, and stores the records for them both in a single destination Kafka topic. The snapshot records that it captures directly from a table are emitted as READ operations. Meanwhile, as users continue to update records in the data collection, and the transaction log is updated to reflect each commit, Debezium emits UPDATE or DELETE operations for each change.

As the snapshot window opens, and Debezium begins processing a snapshot chunk, it delivers snapshot records to a memory buffer. During the snapshot windows, the primary keys of the READ events in the buffer are compared to the primary keys of the incoming streamed events. If no match is found, the streamed event record is sent directly to Kafka. If Debezium detects a match, it discards the buffered READ event, and writes the streamed record to the destination topic, because the streamed event logically supersede the static snapshot event. After the snapshot window for the chunk closes, the buffer contains only READ events for which no related transaction log events exist. Debezium emits these remaining READ events to the table’s Kafka topic.

The connector repeats the process for each snapshot chunk.

Triggering an incremental snapshot

Currently, the only way to initiate an incremental snapshot is to send an ad hoc snapshot signal to the signaling table on the source database.

You submit a signal to the signaling table as SQL INSERT queries.

After Debezium detects the change in the signaling table, it reads the signal, and runs the requested snapshot operation.

The query that you submit specifies the tables to include in the snapshot, and, optionally, specifies the kind of snapshot operation. Currently, the only valid option for snapshots operations is the default value, incremental.

To specify the tables to include in the snapshot, provide a data-collections array that lists the tables or an array of regular expressions used to match tables, for example,

{"data-collections": ["public.MyFirstTable", "public.MySecondTable"]}

The data-collections array for an incremental snapshot signal has no default value. If the data-collections array is empty, Debezium detects that no action is required and does not perform a snapshot.

Prerequisites

• Signaling is enabled.

  • A signaling data collection exists on the source database.

  • The signaling data collection is specified in the signal.data.collection property.

Procedure

1. Send a SQL query to add the ad hoc incremental snapshot request to the signaling table:

   INSERT INTO <signalTable> (id, type, data) VALUES ('<id>', '<snapshotType>', '{"data-collections": ["<tableName>","<tableName>"],"type":"<snapshotType>","additional-condition":"<additional-condition>"}');

   For example,

   INSERT INTO myschema.debezium_signal (id, type, data) (1)
   values ('ad-hoc-1',   (2)
       'execute-snapshot',  (3)
       '{"data-collections": ["schema1.table1", "schema2.table2"], (4)
       "type":"incremental"}, (5)
       "additional-condition":"color=blue"}'); (6)

   The values of the id,type, and data parameters in the command correspond to the fields of the signaling table.

   The following table describes the parameters in the example:

   Table 2. Descriptions of fields in a SQL command for sending an incremental snapshot signal to the signaling table

   Item	Value	Description
   1	myschema.debezium_signal	Specifies the fully-qualified name of the signaling table on the source database.
   2	ad-hoc-1	The id parameter specifies an arbitrary string that is assigned as the id identifier for the signal request. Use this string to identify logging messages to entries in the signaling table. Debezium does not use this string. Rather, during the snapshot, Debezium generates its own id string as a watermarking signal.
   3	execute-snapshot	The type parameter specifies the operation that the signal is intended to trigger.
   4	data-collections	A required component of the data field of a signal that specifies an array of table names or regular expressions to match table names to include in the snapshot. The array lists regular expressions which match tables by their fully-qualified names, using the same format as you use to specify the name of the connector’s signaling table in the signal.data.collection configuration property.
   5	incremental	An optional type component of the data field of a signal that specifies the kind of snapshot operation to run. Currently, the only valid option is the default value, incremental. If you do not specify a value, the connector runs an incremental snapshot.
   6	additional-condition	An optional string, which specifies a condition based on the column(s) of the table(s), to capture a subset of the contents of the tables. For more information about the additional-condition parameter, see Ad hoc incremental snapshots with additional-condition.

Ad hoc incremental snapshots with additional-condition

If you want a snapshot to include only a subset of the content in a table, you can modify the signal request by appending an additional-condition parameter to the snapshot signal.

The SQL query for a typical snapshot takes the following form:

SELECT * FROM <tableName> ....

By adding an additional-condition parameter, you append a WHERE condition to the SQL query, as in the following example:

SELECT * FROM <tableName> WHERE <additional-condition> ....

The following example shows a SQL query to send an ad hoc incremental snapshot request with an additional condition to the signaling table:

INSERT INTO <signalTable> (id, type, data) VALUES ('<id>', '<snapshotType>', '{"data-collections": ["<tableName>","<tableName>"],"type":"<snapshotType>","additional-condition":"<additional-condition>"}');

For example, suppose you have a products table that contains the following columns:

• id (primary key)

• color

• quantity

If you want an incremental snapshot of the products table to include only the data items where color=blue, you can use the following SQL statement to trigger the snapshot:

INSERT INTO myschema.debezium_signal (id, type, data) VALUES('ad-hoc-1', 'execute-snapshot', '{"data-collections": ["schema1.products"],"type":"incremental", "additional-condition":"color=blue"}');

The additional-condition parameter also enables you to pass conditions that are based on more than one column. For example, using the products table from the previous example, you can submit a query that triggers an incremental snapshot that includes the data of only those items for which color=blue and quantity>10:

INSERT INTO myschema.debezium_signal (id, type, data) VALUES('ad-hoc-1', 'execute-snapshot', '{"data-collections": ["schema1.products"],"type":"incremental", "additional-condition":"color=blue AND quantity>10"}');

The following example, shows the JSON for an incremental snapshot event that is captured by a connector.

Example: Incremental snapshot event message

{
    "before":null,
    "after": {
        "pk":"1",
        "value":"New data"
    },
    "source": {
        ...
        "snapshot":"incremental" (1)
    },
    "op":"r", (2)
    "ts_ms":"1620393591654",
    "transaction":null
}

Stopping an incremental snapshot

You can also stop an incremental snapshot by sending a signal to the table on the source database. You submit a stop snapshot signal to the table by sending a SQL INSERT query.

After Debezium detects the change in the signaling table, it reads the signal, and stops the incremental snapshot operation if it’s in progress.

The query that you submit specifies the snapshot operation of incremental, and, optionally, the tables of the current running snapshot to be removed.

Prerequisites

• Signaling is enabled.

  • A signaling data collection exists on the source database.

  • The signaling data collection is specified in the signal.data.collection property.

Procedure

1. Send a SQL query to stop the ad hoc incremental snapshot to the signaling table:

   INSERT INTO <signalTable> (id, type, data) values ('<id>', 'stop-snapshot', '{"data-collections": ["<tableName>","<tableName>"],"type":"incremental"}');

   For example,

   INSERT INTO myschema.debezium_signal (id, type, data) (1)
   values ('ad-hoc-1',   (2)
       'stop-snapshot',  (3)
       '{"data-collections": ["schema1.table1", "schema2.table2"], (4)
       "type":"incremental"}'); (5)

   The values of the id, type, and data parameters in the signal command correspond to the fields of the signaling table.

   The following table describes the parameters in the example:

   Table 3. Descriptions of fields in a SQL command for sending a stop incremental snapshot signal to the signaling table

   Item	Value	Description
   1	myschema.debezium_signal	Specifies the fully-qualified name of the signaling table on the source database.
   2	ad-hoc-1	The id parameter specifies an arbitrary string that is assigned as the id identifier for the signal request. Use this string to identify logging messages to entries in the signaling table. Debezium does not use this string.
   3	stop-snapshot	Specifies type parameter specifies the operation that the signal is intended to trigger.
   4	data-collections	An optional component of the data field of a signal that specifies an array of table names or regular expressions to match table names to remove from the snapshot. The array lists regular expressions which match tables by their fully-qualified names, using the same format as you use to specify the name of the connector’s signaling table in the signal.data.collection configuration property. If this component of the data field is omitted, the signal stops the entire incremental snapshot that is in progress.
   5	incremental	A required component of the data field of a signal that specifies the kind of snapshot operation that is to be stopped. Currently, the only valid option is incremental. If you do not specify a type value, the signal fails to stop the incremental snapshot.

Reading the change data tables

When the connector first starts, it takes a structural snapshot of the structure of the captured tables and persists this information to its internal database schema history topic. The connector then identifies a change table for each source table, and completes the following steps.

1. For each change table, the connector read all of the changes that were created between the last stored maximum LSN and the current maximum LSN.

2. The connector sorts the changes that it reads in ascending order, based on the values of their commit LSN and change LSN. This sorting order ensures that the changes are replayed by Debezium in the same order in which they occurred in the database.

3. The connector passes the commit and change LSNs as offsets to Kafka Connect.

4. The connector stores the maximum LSN and restarts the process from Step 1.

After a restart, the connector resumes processing from the last offset (commit and change LSNs) that it read.

The connector is able to detect whether CDC is enabled or disabled for included source tables and adjust its behavior.

No maximum LSN recorded in the database

There may be situations when no maximum LSN is recorded in the database because:

1. SQL Server Agent is not running

2. No changes are recorded in the change table yet

3. Database has low activity and the cdc clean up job periodically clears entries from the cdc tables

Out of these possibilities, since a running SQL Server Agent is a prerequisite, No 1. is a real problem (while No 2. and 3. are normal).

In order to mitigate this issue and differentiate between No 1. and the others, a check for the status of the SQL Server Agent is done through the following query "SELECT CASE WHEN dss.[status]=4 THEN 1 ELSE 0 END AS isRunning FROM [#db].sys.dm_server_services dss WHERE dss.[servicename] LIKE N’SQL Server Agent (%';". If the SQL Server Agent is not running, an ERROR is written in the log: “No maximum LSN recorded in the database; SQL Server Agent is not running”.

Limitations

SQL Server specifically requires the base object to be a table in order to create a change capture instance. As consequence, capturing changes from indexed views (aka. materialized views) is not supported by SQL Server and hence Debezium SQL Server connector.

Topic names

By default, the SQL Server connector writes events for all INSERT, UPDATE, and DELETE operations that occur in a table to a single Apache Kafka topic that is specific to that table. The connector uses the following convention to name change event topics: _<topicPrefix>_._<schemaName>_._<tableName>_

The following list provides definitions for the components of the default name:

topicPrefix

The logical name of the server, as specified by the topic.prefix configuration property.

schemaName

The name of the database schema in which the change event occurred.

tableName

The name of the database table in which the change event occurred.

For example, if fulfillment is the logical server name, and dbo is the schema name, and the database contains tables with the names products, products_on_hand, customers, and orders, the connector would stream change event records to the following Kafka topics:

• fulfillment.testDB.dbo.products

• fulfillment.testDB.dbo.products_on_hand

• fulfillment.testDB.dbo.customers

• fulfillment.testDB.dbo.orders

The connector applies similar naming conventions to label its internal database schema history topics, schema change topics, and transaction metadata topics.

If the default topic name do not meet your requirements, you can configure custom topic names. To configure custom topic names, you specify regular expressions in the logical topic routing SMT. For more information about using the logical topic routing SMT to customize topic naming, see Topic routing.

Schema change topic

For each table for which CDC is enabled, the Debezium SQL Server connector stores a history of the schema change events that are applied to captured tables in the database. The connector writes schema change events to a Kafka topic named _<topicPrefix>_, where _topicPrefix_ is the logical server name that is specified in the topic.prefix configuration property.

Messages that the connector sends to the schema change topic contain a payload, and, optionally, also contain the schema of the change event message. The payload of a schema change event message includes the following elements:

databaseName

The name of the database to which the statements are applied. The value of databaseName serves as the message key.

tableChanges

A structured representation of the entire table schema after the schema change. The tableChanges field contains an array that includes entries for each column of the table. Because the structured representation presents data in JSON or Avro format, consumers can easily read messages without first processing them through a DDL parser.

Debezium emits a message to the schema change topic when the following events occur:

• You enable CDC for a table.

• You disable CDC for a table.

• You alter the structure of a table for which CDC is enabled by following the schema evolution procedure.

Example: Message emitted to the SQL Server connector schema change topic

The following example shows a message in the schema change topic. The message contains a logical representation of the table schema.

{
  "schema": {
  ...
  },
  "payload": {
    "source": {
      "version": "2.1.4.Final",
      "connector": "sqlserver",
      "name": "server1",
      "ts_ms": 0,
      "snapshot": "true",
      "db": "testDB",
      "schema": "dbo",
      "table": "customers",
      "change_lsn": null,
      "commit_lsn": "00000025:00000d98:00a2",
      "event_serial_no": null
    },
    "ts_ms": 1588252618953, (1)
    "databaseName": "testDB", (2)
    "schemaName": "dbo",
    "ddl": null, (3)
    "tableChanges": [ (4)
      {
        "type": "CREATE", (5)
        "id": "\"testDB\".\"dbo\".\"customers\"", (6)
        "table": { (7)
          "defaultCharsetName": null,
          "primaryKeyColumnNames": [ (8)
            "id"
          ],
          "columns": [ (9)
            {
              "name": "id",
              "jdbcType": 4,
              "nativeType": null,
              "typeName": "int identity",
              "typeExpression": "int identity",
              "charsetName": null,
              "length": 10,
              "scale": 0,
              "position": 1,
              "optional": false,
              "autoIncremented": false,
              "generated": false
            },
            {
              "name": "first_name",
              "jdbcType": 12,
              "nativeType": null,
              "typeName": "varchar",
              "typeExpression": "varchar",
              "charsetName": null,
              "length": 255,
              "scale": null,
              "position": 2,
              "optional": false,
              "autoIncremented": false,
              "generated": false
            },
            {
              "name": "last_name",
              "jdbcType": 12,
              "nativeType": null,
              "typeName": "varchar",
              "typeExpression": "varchar",
              "charsetName": null,
              "length": 255,
              "scale": null,
              "position": 3,
              "optional": false,
              "autoIncremented": false,
              "generated": false
            },
            {
              "name": "email",
              "jdbcType": 12,
              "nativeType": null,
              "typeName": "varchar",
              "typeExpression": "varchar",
              "charsetName": null,
              "length": 255,
              "scale": null,
              "position": 4,
              "optional": false,
              "autoIncremented": false,
              "generated": false
            }
          ],
          "attributes": [ (10)
            {
              "customAttribute": "attributeValue"
            }
          ]
        }
      }
    ]
  }
}

In messages that the connector sends to the schema change topic, the key is the name of the database that contains the schema change. In the following example, the payload field contains the key:

{
  "schema": {
    "type": "struct",
    "fields": [
      {
        "type": "string",
        "optional": false,
        "field": "databaseName"
      }
    ],
    "optional": false,
    "name": "io.debezium.connector.sqlserver.SchemaChangeKey"
  },
  "payload": {
    "databaseName": "testDB"
  }
}

Data change events

The Debezium SQL Server connector generates a data change event for each row-level INSERT, UPDATE, and DELETE operation. Each event contains a key and a value. The structure of the key and the value depends on the table that was changed.

Debezium and Kafka Connect are designed around continuous streams of event messages. However, the structure of these events may change over time, which can be difficult for consumers to handle. To address this, each event contains the schema for its content or, if you are using a schema registry, a schema ID that a consumer can use to obtain the schema from the registry. This makes each event self-contained.

The following skeleton JSON shows the basic four parts of a change event. However, how you configure the Kafka Connect converter that you choose to use in your application determines the representation of these four parts in change events. A schema field is in a change event only when you configure the converter to produce it. Likewise, the event key and event payload are in a change event only if you configure a converter to produce it. If you use the JSON converter and you configure it to produce all four basic change event parts, change events have this structure:

{
 "schema": { (1)
   ...
  },
 "payload": { (2)
   ...
 },
 "schema": { (3)
   ...
 },
 "payload": { (4)
   ...
 },
}

By default, the connector streams change event records to topics with names that are the same as the event’s originating table. For more information, see topic names.

Change event keys

A change event’s key contains the schema for the changed table’s key and the changed row’s actual key. Both the schema and its corresponding payload contain a field for each column in the changed table’s primary key (or unique key constraint) at the time the connector created the event.

Consider the following customers table, which is followed by an example of a change event key for this table.

Example table

CREATE TABLE customers (
  id INTEGER IDENTITY(1001,1) NOT NULL PRIMARY KEY,
  first_name VARCHAR(255) NOT NULL,
  last_name VARCHAR(255) NOT NULL,
  email VARCHAR(255) NOT NULL UNIQUE
);

Example change event key

Every change event that captures a change to the customers table has the same event key schema. For as long as the customers table has the previous definition, every change event that captures a change to the customers table has the following key structure, which in JSON, looks like this:

{
    "schema": { (1)
        "type": "struct",
        "fields": [ (2)
            {
                "type": "int32",
                "optional": false,
                "field": "id"
            }
        ],
        "optional": false, (3)
        "name": "server1.testDB.dbo.customers.Key" (4)
    },
    "payload": { (5)
        "id": 1004
    }
}

Change event values

The value in a change event is a bit more complicated than the key. Like the key, the value has a schema section and a payload section. The schema section contains the schema that describes the Envelope structure of the payload section, including its nested fields. Change events for operations that create, update or delete data all have a value payload with an envelope structure.

Consider the same sample table that was used to show an example of a change event key:

CREATE TABLE customers (
  id INTEGER IDENTITY(1001,1) NOT NULL PRIMARY KEY,
  first_name VARCHAR(255) NOT NULL,
  last_name VARCHAR(255) NOT NULL,
  email VARCHAR(255) NOT NULL UNIQUE
);

The value portion of a change event for a change to this table is described for each event type.

create events

The following example shows the value portion of a change event that the connector generates for an operation that creates data in the customers table:

{
  "schema": { (1)
    "type": "struct",
    "fields": [
      {
        "type": "struct",
        "fields": [
          {
            "type": "int32",
            "optional": false,
            "field": "id"
          },
          {
            "type": "string",
            "optional": false,
            "field": "first_name"
          },
          {
            "type": "string",
            "optional": false,
            "field": "last_name"
          },
          {
            "type": "string",
            "optional": false,
            "field": "email"
          }
        ],
        "optional": true,
        "name": "server1.dbo.testDB.customers.Value", (2)
        "field": "before"
      },
      {
        "type": "struct",
        "fields": [
          {
            "type": "int32",
            "optional": false,
            "field": "id"
          },
          {
            "type": "string",
            "optional": false,
            "field": "first_name"
          },
          {
            "type": "string",
            "optional": false,
            "field": "last_name"
          },
          {
            "type": "string",
            "optional": false,
            "field": "email"
          }
        ],
        "optional": true,
        "name": "server1.dbo.testDB.customers.Value",
        "field": "after"
      },
      {
        "type": "struct",
        "fields": [
          {
            "type": "string",
            "optional": false,
            "field": "version"
          },
          {
            "type": "string",
            "optional": false,
            "field": "connector"
          },
          {
            "type": "string",
            "optional": false,
            "field": "name"
          },
          {
            "type": "int64",
            "optional": false,
            "field": "ts_ms"
          },
          {
            "type": "boolean",
            "optional": true,
            "default": false,
            "field": "snapshot"
          },
          {
            "type": "string",
            "optional": false,
            "field": "db"
          },
          {
            "type": "string",
            "optional": false,
            "field": "schema"
          },
          {
            "type": "string",
            "optional": false,
            "field": "table"
          },
          {
            "type": "string",
            "optional": true,
            "field": "change_lsn"
          },
          {
            "type": "string",
            "optional": true,
            "field": "commit_lsn"
          },
          {
            "type": "int64",
            "optional": true,
            "field": "event_serial_no"
          }
        ],
        "optional": false,
        "name": "io.debezium.connector.sqlserver.Source", (3)
        "field": "source"
      },
      {
        "type": "string",
        "optional": false,
        "field": "op"
      },
      {
        "type": "int64",
        "optional": true,
        "field": "ts_ms"
      }
    ],
    "optional": false,
    "name": "server1.dbo.testDB.customers.Envelope" (4)
  },
  "payload": { (5)
    "before": null, (6)
    "after": { (7)
      "id": 1005,
      "first_name": "john",
      "last_name": "doe",
      "email": "john.doe@example.org"
    },
    "source": { (8)
      "version": "2.1.4.Final",
      "connector": "sqlserver",
      "name": "server1",
      "ts_ms": 1559729468470,
      "snapshot": false,
      "db": "testDB",
      "schema": "dbo",
      "table": "customers",
      "change_lsn": "00000027:00000758:0003",
      "commit_lsn": "00000027:00000758:0005",
      "event_serial_no": "1"
    },
    "op": "c", (9)
    "ts_ms": 1559729471739 (10)
  }
}

update events

The value of a change event for an update in the sample customers table has the same schema as a create event for that table. Likewise, the event value’s payload has the same structure. However, the event value payload contains different values in an update event. Here is an example of a change event value in an event that the connector generates for an update in the customers table:

{
  "schema": { ... },
  "payload": {
    "before": { (1)
      "id": 1005,
      "first_name": "john",
      "last_name": "doe",
      "email": "john.doe@example.org"
    },
    "after": { (2)
      "id": 1005,
      "first_name": "john",
      "last_name": "doe",
      "email": "noreply@example.org"
    },
    "source": { (3)
      "version": "2.1.4.Final",
      "connector": "sqlserver",
      "name": "server1",
      "ts_ms": 1559729995937,
      "snapshot": false,
      "db": "testDB",
      "schema": "dbo",
      "table": "customers",
      "change_lsn": "00000027:00000ac0:0002",
      "commit_lsn": "00000027:00000ac0:0007",
      "event_serial_no": "2"
    },
    "op": "u", (4)
    "ts_ms": 1559729998706  (5)
  }
}

delete events

The value in a delete change event has the same schema portion as create and update events for the same table. The payload portion in a delete event for the sample customers table looks like this:

{
  "schema": { ... },
  },
  "payload": {
    "before": { <>
      "id": 1005,
      "first_name": "john",
      "last_name": "doe",
      "email": "noreply@example.org"
    },
    "after": null, (2)
    "source": { (3)
      "version": "2.1.4.Final",
      "connector": "sqlserver",
      "name": "server1",
      "ts_ms": 1559730445243,
      "snapshot": false,
      "db": "testDB",
      "schema": "dbo",
      "table": "customers",
      "change_lsn": "00000027:00000db0:0005",
      "commit_lsn": "00000027:00000db0:0007",
      "event_serial_no": "1"
    },
    "op": "d", (4)
    "ts_ms": 1559730450205 (5)
  }
}

SQL Server connector events are designed to work with Kafka log compaction. Log compaction enables removal of some older messages as long as at least the most recent message for every key is kept. This lets Kafka reclaim storage space while ensuring that the topic contains a complete data set and can be used for reloading key-based state.

Tombstone events

When a row is deleted, the delete event value still works with log compaction, because Kafka can remove all earlier messages that have that same key. However, for Kafka to remove all messages that have that same key, the message value must be null. To make this possible, after Debezium’s SQL Server connector emits a delete event, the connector emits a special tombstone event that has the same key but a null value.

Transaction metadata

Debezium can generate events that represent transaction boundaries and that enrich data change event messages.

Database transactions are represented by a statement block that is enclosed between the BEGIN and END keywords. Debezium generates transaction boundary events for the BEGIN and END delimiters in every transaction. Transaction boundary events contain the following fields:

status

BEGIN or END.

id

String representation of the unique transaction identifier.

ts_ms

The time of a transaction boundary event (BEGIN or END event) at the data source. If the data source does not provide Debezium with the event time, then the field instead represents the time at which Debezium processes the event.

event_count (for END events)

Total number of events emmitted by the transaction.

data_collections (for END events)

An array of pairs of data_collection and event_count elements that indicates the number of events that the connector emits for changes that originate from a data collection.

The following example shows a typical transaction boundary message:

Example: SQL Server connector transaction boundary event

{
  "status": "BEGIN",
  "id": "00000025:00000d08:0025",
  "ts_ms": 1486500577125,
  "event_count": null,
  "data_collections": null
}
{
  "status": "END",
  "id": "00000025:00000d08:0025",
  "ts_ms": 1486500577691,
  "event_count": 2,
  "data_collections": [
    {
      "data_collection": "testDB.dbo.testDB.tablea",
      "event_count": 1
    },
    {
      "data_collection": "testDB.dbo.testDB.tableb",
      "event_count": 1
    }
  ]
}

Unless overridden via the topic.transaction option, transaction events are written to the topic named .transaction.

Change data event enrichment

When transaction metadata is enabled, the data message Envelope is enriched with a new transaction field. This field provides information about every event in the form of a composite of fields:

id

String representation of unique transaction identifier

total_order

The absolute position of the event among all events generated by the transaction

data_collection_order

The per-data collection position of the event among all events that were emitted by the transaction

The following example shows what a typical message looks like:

{
  "before": null,
  "after": {
    "pk": "2",
    "aa": "1"
  },
  "source": {
...
  },
  "op": "c",
  "ts_ms": "1580390884335",
  "transaction": {
    "id": "00000025:00000d08:0025",
    "total_order": "1",
    "data_collection_order": "1"
  }
}

Data type mappings

The Debezium SQL Server connector represents changes to table row data by producing events that are structured like the table in which the row exists. Each event contains fields to represent the column values for the row. The way in which an event represents the column values for an operation depends on the SQL data type of the column. In the event, the connector maps the fields for each SQL Server data type to both a literal type and a semantic type.

The connector can map SQL Server data types to both literal and semantic types.

Literal type

Describes how the value is literally represented by using Kafka Connect schema types, namely INT8, INT16, INT32, INT64, FLOAT32, FLOAT64, BOOLEAN, STRING, BYTES, ARRAY, MAP, and STRUCT.

Semantic type

Describes how the Kafka Connect schema captures the meaning of the field using the name of the Kafka Connect schema for the field.

If the default data type conversions do not meet your needs, you can create a custom converter for the connector.

Basic types

The following table shows how the connector maps basic SQL Server data types.

Other data type mappings are described in the following sections.

If present, a column’s default value is propagated to the corresponding field’s Kafka Connect schema. Change messages will contain the field’s default value (unless an explicit column value had been given), so there should rarely be the need to obtain the default value from the schema. Passing the default value helps though with satisfying the compatibility rules when using Avro as serialization format together with the Confluent schema registry.

Temporal values

Other than SQL Server’s DATETIMEOFFSET data type (which contain time zone information), the other temporal types depend on the value of the time.precision.mode configuration property. When the time.precision.mode configuration property is set to adaptive (the default), then the connector will determine the literal type and semantic type for the temporal types based on the column’s data type definition so that events exactly represent the values in the database:

When the time.precision.mode configuration property is set to connect, then the connector will use the predefined Kafka Connect logical types. This may be useful when consumers only know about the built-in Kafka Connect logical types and are unable to handle variable-precision time values. On the other hand, since SQL Server supports tenth of microsecond precision, the events generated by a connector with the connect time precision mode will result in a loss of precision when the database column has a fractional second precision value greater than 3:

Timestamp values

The DATETIME, SMALLDATETIME and DATETIME2 types represent a timestamp without time zone information. Such columns are converted into an equivalent Kafka Connect value based on UTC. So for instance the DATETIME2 value “2018-06-20 15:13:16.945104” is represented by a io.debezium.time.MicroTimestamp with the value “1529507596945104”.

Note that the timezone of the JVM running Kafka Connect and Debezium does not affect this conversion.

Decimal values

Debezium connectors handle decimals according to the setting of the decimal.handling.mode connector configuration property.

decimal.handling.mode=precise

decimal.handling.mode=double

decimal.handling.mode=string

Setting up SQL Server

For Debezium to capture change events from SQL Server tables, a SQL Server administrator with the necessary privileges must first run a query to enable CDC on the database. The administrator must then enable CDC for each table that you want Debezium to capture.

After CDC is applied, it captures all of the INSERT, UPDATE, and DELETE operations that are committed to the tables for which CDD is enabled. The Debezium connector can then capture these events and emit them to Kafka topics.

Enabling CDC on the SQL Server database

Before you can enable CDC for a table, you must enable it for the SQL Server database. A SQL Server administrator enables CDC by running a system stored procedure. System stored procedures can be run by using SQL Server Management Studio, or by using Transact-SQL.

Prerequisites

• You are a member of the sysadmin fixed server role for the SQL Server.

• You are a db_owner of the database.

• The SQL Server Agent is running.

Procedure

1. From the View menu in SQL Server Management Studio, click Template Explorer.

2. In the Template Browser, expand SQL Server Templates.

3. Expand Change Data Capture > Configuration and then click Enable Database for CDC.

4. In the template, replace the database name in the USE statement with the name of the database that you want to enable for CDC.

5. Run the stored procedure sys.sp_cdc_enable_db to enable the database for CDC.

   After the database is enabled for CDC, a schema with the name cdc is created, along with a CDC user, metadata tables, and other system objects.

   The following example shows how to enable CDC for the database MyDB:

   Example: Enabling a SQL Server database for the CDC template

   USE MyDB
   GO
   EXEC sys.sp_cdc_enable_db
   GO

Enabling CDC on a SQL Server table

A SQL Server administrator must enable change data capture on the source tables that you want to Debezium to capture. The database must already be enabled for CDC. To enable CDC on a table, a SQL Server administrator runs the stored procedure sys.sp_cdc_enable_table for the table. The stored procedures can be run by using SQL Server Management Studio, or by using Transact-SQL. SQL Server CDC must be enabled for every table that you want to capture.

Prerequisites

• CDC is enabled on the SQL Server database.

• The SQL Server Agent is running.

• You are a member of the db_owner fixed database role for the database.

Procedure

1. From the View menu in SQL Server Management Studio, click Template Explorer.

2. In the Template Browser, expand SQL Server Templates.

3. Expand Change Data Capture > Configuration, and then click Enable Table Specifying Filegroup Option.

4. In the template, replace the table name in the USE statement with the name of the table that you want to capture.

5. Run the stored procedure sys.sp_cdc_enable_table.

   The following example shows how to enable CDC for the table MyTable:

   Example: Enabling CDC for a SQL Server table

   USE MyDB
   GO
   EXEC sys.sp_cdc_enable_table
   @source_schema = N'dbo',
   @source_name   = N'MyTable', (1)
   @role_name     = N'MyRole',  (2)
   @filegroup_name = N'MyDB_CT',(3)
   @supports_net_changes = 0
   GO

   1	Specifies the name of the table that you want to capture.
   2	Specifies a role MyRole to which you can add users to whom you want to grant SELECT permission on the captured columns of the source table. Users in the sysadmin or db_owner role also have access to the specified change tables. Set the value of @role_name to NULL, to allow only members in the sysadmin or db_owner to have full access to captured information.
   3	Specifies the filegroup where SQL Server places the change table for the captured table. The named filegroup must already exist. It is best not to locate change tables in the same filegroup that you use for source tables.

Verifying that the user has access to the CDC table

A SQL Server administrator can run a system stored procedure to query a database or table to retrieve its CDC configuration information. The stored procedures can be run by using SQL Server Management Studio, or by using Transact-SQL.

Prerequisites

• You have SELECT permission on all of the captured columns of the capture instance. Members of the db_owner database role can view information for all of the defined capture instances.

• You have membership in any gating roles that are defined for the table information that the query includes.

Procedure

1. From the View menu in SQL Server Management Studio, click Object Explorer.

2. From the Object Explorer, expand Databases, and then expand your database object, for example, MyDB.

3. Expand Programmability > Stored Procedures > System Stored Procedures.

4. Run the sys.sp_cdc_help_change_data_capture stored procedure to query the table.

   Queries should not return empty results.

   The following example runs the stored procedure sys.sp_cdc_help_change_data_capture on the database MyDB:

   Example: Querying a table for CDC configuration information

   USE MyDB;
   GO
   EXEC sys.sp_cdc_help_change_data_capture
   GO

   The query returns configuration information for each table in the database that is enabled for CDC and that contains change data that the caller is authorized to access. If the result is empty, verify that the user has privileges to access both the capture instance and the CDC tables.

SQL Server on Azure

The Debezium SQL Server connector can be used with SQL Server on Azure. Refer to this example for configuring CDC for SQL Server on Azure and using it with Debezium.

SQL Server Always On

The SQL Server connector can capture changes from an Always On read-only replica.

Prerequisites

• Change data capture is configured and enabled on the primary node. SQL Server does not support CDC directly on replicas.

• The configuration option database.applicationIntent is set to ReadOnly. This is required by SQL Server. When Debezium detects this configuration option, it responds by taking the following actions:

  • Sets snapshot.isolation.mode to snapshot, which is the only one transaction isolation mode supported for read-only replicas.

  • Commits the (read-only) transaction in every execution of the streaming query loop, which is necessary to get the latest view of CDC data.

Effect of SQL Server capture job agent configuration on server load and latency

When a database administrator enables change data capture for a source table, the capture job agent begins to run. The agent reads new change event records from the transaction log and replicates the event records to a change data table. Between the time that a change is committed in the source table, and the time that the change appears in the corresponding change table, there is always a small latency interval. This latency interval represents a gap between when changes occur in the source table and when they become available for Debezium to stream to Apache Kafka.

Ideally, for applications that must respond quickly to changes in data, you want to maintain close synchronization between the source and change tables. You might imagine that running the capture agent to continuously process change events as rapidly as possible might result in increased throughput and reduced latency — populating change tables with new event records as soon as possible after the events occur, in near real time. However, this is not necessarily the case. There is a performance penalty to pay in the pursuit of more immediate synchronization. Each time that the capture job agent queries the database for new event records, it increases the CPU load on the database host. The additional load on the server can have a negative effect on overall database performance, and potentially reduce transaction efficiency, especially during times of peak database use.

It’s important to monitor database metrics so that you know if the database reaches the point where the server can no longer support the capture agent’s level of activity. If you notice performance problems, there are SQL Server capture agent settings that you can modify to help balance the overall CPU load on the database host with a tolerable degree of latency.

SQL Server capture job agent configuration parameters

On SQL Server, parameters that control the behavior of the capture job agent are defined in the SQL Server table msdb.dbo.cdc_jobs. If you experience performance issues while running the capture job agent, adjust capture jobs settings to reduce CPU load by running the sys.sp_cdc_change_job stored procedure and supplying new values.

The following parameters are the most significant for modifying capture agent behavior for use with the Debezium SQL Server connector:

pollinginterval

• Specifies the number of seconds that the capture agent waits between log scan cycles.

• A higher value reduces the load on the database host and increases latency.

• A value of 0 specifies no wait between scans.

• The default value is 5.

maxtrans

• Specifies the maximum number of transactions to process during each log scan cycle. After the capture job processes the specified number of transactions, it pauses for the length of time that the pollinginterval specifies before the next scan begins.

• A lower value reduces the load on the database host and increases latency.

• The default value is 500.

maxscans

• Specifies a limit on the number of scan cycles that the capture job can attempt in capturing the full contents of the database transaction log. If the continuous parameter is set to 1, the job pauses for the length of time that the pollinginterval specifies before it resumes scanning.

• A lower values reduces the load on the database host and increases latency.

• The default value is 10.

Additional resources

• For more information about capture agent parameters, see the SQL Server documentation.

Deployment

To deploy a Debezium SQL Server connector, you install the Debezium SQL Server connector archive, configure the connector, and start the connector by adding its configuration to Kafka Connect.

Prerequisites

• Apache ZooKeeper, Apache Kafka, and Kafka Connect are installed.

• SQL Server is installed, is configured for CDC, and is ready to be used with the Debezium connector.

Procedure

1. Download the Debezium SQL Server connector plug-in archive

2. Extract the files into your Kafka Connect environment.

3. Add the directory with the JAR files to Kafka Connect’s plugin.path.

4. Configure the connector and add the configuration to your Kafka Connect cluster.

5. Restart your Kafka Connect process to pick up the new JAR files.

If you are working with immutable containers, see Debezium’s container images for Åpache ZooKeeper, Apache Kafka, and Kafka Connect. You can pull the official container images for Microsoft SQL Server on Linux from Docker Hub.

You can also run Debezium on Kubernetes and OpenShift.

SQL Server connector configuration example

Following is an example of the configuration for a connector instance that captures data from a SQL Server server at port 1433 on 192.168.99.100, which we logically name fullfillment. Typically, you configure the Debezium SQL Server connector in a JSON file by setting the configuration properties that are available for the connector.

You can choose to produce events for a subset of the schemas and tables in a database. Optionally, you can ignore, mask, or truncate columns that contain sensitive data, that are larger than a specified size, or that you do not need.

{
    "name": "inventory-connector", (1)
    "config": {
        "connector.class": "io.debezium.connector.sqlserver.SqlServerConnector", (2)
        "database.hostname": "192.168.99.100", (3)
        "database.port": "1433", (4)
        "database.user": "sa", (5)
        "database.password": "Password!", (6)
        "database.names": "testDB1,testDB2", (7)
        "topic.prefix": "fullfillment", (8)
        "table.include.list": "dbo.customers", (9)
        "schema.history.internal.kafka.bootstrap.servers": "kafka:9092", (10)
        "schema.history.internal.kafka.topic": "schemahistory.fullfillment" (11)
        "database.ssl.truststore": "path/to/trust-store" (12)
        "database.ssl.truststore.password": "password-for-trust-store" (13)
    }
}

For the complete list of the configuration properties that you can set for the Debezium SQL Server connector, see SQL Server connector properties.

You can send this configuration with a POST command to a running Kafka Connect service. The service records the configuration and start up the one connector task that performs the following tasks:

• Connects to the SQL Server database.

• Reads the transaction log.

• Records change events to Kafka topics.

Adding connector configuration

To start running a Debezium SQL Server connector, create a connector configuration, and add the configuration to your Kafka Connect cluster.

Prerequisites

• CDC is enabled on SQL Server.

• The Debezium SQL Server connector is installed.

Procedure

1. Create a configuration for the SQL Server connector.

2. Use the Kafka Connect REST API to add that connector configuration to your Kafka Connect cluster.

Results

When the connector starts, it performs a consistent snapshot of the SQL Server databases that the connector is configured for. The connector then starts generating data change events for row-level operations and streaming the change event records to Kafka topics.

Connector properties

The Debezium SQL Server connector has numerous configuration properties that you can use to achieve the right connector behavior for your application. Many properties have default values.

Information about the properties is organized as follows:

• Required connector configuration properties

• Advanced connector configuration properties

• Database schema history connector configuration properties that control how Debezium processes events that it reads from the database schema history topic.

  • Pass-through database schema history properties

• Pass-through database driver properties that control the behavior of the database driver.

Required Debezium SQL Server connector configuration properties

The following configuration properties are required unless a default value is available.

column.mask.hash.SHA-256.with.salt.CzQMA0cB5K = inventory.orders.customerName, inventory.shipment.customerName

Advanced SQL Server connector configuration properties

The following advanced configuration properties have good defaults that will work in most situations and therefore rarely need to be specified in the connector’s configuration.

isbn.type: io.debezium.test.IsbnConverter

isbn.schema.name: io.debezium.sqlserver.type.Isbn

“snapshot.select.statement.overrides”: “customer.orders”,
“snapshot.select.statement.overrides.customer.orders”: “SELECT * FROM [customers].[orders] WHERE delete_flag = 0 ORDER BY id DESC”

Debezium SQL Server connector database schema history configuration properties

Debezium provides a set of schema.history.internal.* properties that control how the connector interacts with the schema history topic.

The following table describes the schema.history.internal properties for configuring the Debezium connector.

Pass-through database schema history properties for configuring producer and consumer clients

Debezium relies on a Kafka producer to write schema changes to database schema history topics. Similarly, it relies on a Kafka consumer to read from database schema history topics when a connector starts. You define the configuration for the Kafka producer and consumer clients by assigning values to a set of pass-through configuration properties that begin with the schema.history.internal.producer.* and schema.history.internal.consumer.* prefixes. The pass-through producer and consumer database schema history properties control a range of behaviors, such as how these clients secure connections with the Kafka broker, as shown in the following example:

schema.history.internal.producer.security.protocol=SSL
schema.history.internal.producer.ssl.keystore.location=/var/private/ssl/kafka.server.keystore.jks
schema.history.internal.producer.ssl.keystore.password=test1234
schema.history.internal.producer.ssl.truststore.location=/var/private/ssl/kafka.server.truststore.jks
schema.history.internal.producer.ssl.truststore.password=test1234
schema.history.internal.producer.ssl.key.password=test1234
schema.history.internal.consumer.security.protocol=SSL
schema.history.internal.consumer.ssl.keystore.location=/var/private/ssl/kafka.server.keystore.jks
schema.history.internal.consumer.ssl.keystore.password=test1234
schema.history.internal.consumer.ssl.truststore.location=/var/private/ssl/kafka.server.truststore.jks
schema.history.internal.consumer.ssl.truststore.password=test1234
schema.history.internal.consumer.ssl.key.password=test1234