Setting Up CDC

Describes the requirements and how to set up Change Data Capture (CDC).

To set up the Change Data Capture (CDC) feature, the following must exist or be created:
  • MapR Database source table (JSON or binary)
  • MapR Event Store For Apache Kafka changelog stream
  • MapR Event Store For Apache Kafka stream topic
  • MapR Database table changelog relationship between the source table and the destination stream topic

Before Setting Up CDC

The destination MapR Event Store For Apache Kafka stream can be in the same cluster as the MapR Database source table or it can be on a remote MapR cluster. Regardless of where the stream is located, propagating changed data requires a gateway.

Typically, gateways are setup by installing the gateway on the destination cluster and specifying the gateway node(s) on the source cluster. However, if the stream and the HPE Ezmeral Data Fabric Database source table are in the same cluster, install the gateway on that cluster and specify the gateway nodes(s) in the cluster. See Administering MapR Gateways and Configuring Gateways for Table and Stream Replication.

The following diagram shows a simple CDC data model with one source table to one destination topic on one stream; however, more complex CDC scenarios can be implemented with multiple gateways. See Data Modeling and CDC.

IMPORTANT If you have a secure cluster, secure configuration must be setup. See Configuring Secure Clusters for Cross-Cluster Mirroring and Replication.

Create Table

An MapR Database table (JSON or binary) must be established for the CDC data. You can create a new table and add data or use an existing table with data. See maprcli table create for creating a new table or use the Control System. Example code is provided for completing this task using either the maprcli or REST. Alternatively, depending on whether you are establishing JSON documents or binary files, you can use the following:
ATTENTION Ensure that a volume exists and mounted for both tables and streams. Even though MapR Database tables and MapR Event Store For Apache Kafka stream can exist in the same volume, for organizational purposes, you could create separate volumes for both tables and streams.
The following code examples show how to:
  • Create and mount a volume for a source table.
  • Create a new binary table. The -tabletype parameter's default setting is binary so you don't need to specify this parameter.
  • Create a new JSON table.
// Create Volume for table
maprcli volume create -name tableVolume -path /tableVolume
                    
// Create Binary table
maprcli table create -path /tableVolume/cdcTable

// Create JSON table
maprcli table create -path /tableVolume/cdcTable -tabletype json
// Create Volume for table
https://10.10.100.17:8443/rest/volume/create?name=tableVolume&path=/tableVolume

// Create Binary table
https://10.10.100.17:8443/rest/table/create?path=/tableVolume/cdcTable

// Create JSON table
https://10.10.100.17:8443/rest/table/create?path=/tableVolume/cdcTable&tabletype=json

Create Stream

An MapR Event Store For Apache Kafka changelog stream must be created for the propagated changed data records using the maprcli stream create -ischangelog parameter. See maprcli stream create or use the Control System.

NOTE Ensure that a volume exists and mounted for both tables and streams. Even though MapR Database tables and MapR Event Store For Apache Kafka stream can exist in the same volume, for organizational purposes, you could create separate volumes for tables and streams.
IMPORTANT The changelog stream's default partitions can impact how many partitions a stream topic can have. This is because once you create a stream topic for a changelog stream, the number of topic partitions is locked. The number of topic partitions cannot change.
  • If the stream topic create command is used to create a stream topic, then the number of topic partitions can be set at creation time and then is locked.
  • If the table changelog add command is used to add a stream topic (as well as establish a relationship between the source table and the changelog stream), then the number of topic partitions is inherited from the changelog stream and is locked.
The following code examples show how to:
  • Create and mount a volume for a changelog stream.
  • Create a changelog stream using the default partitions value of one (1).
  • Create a changelog stream changing the default partitions to three (3).
// Create Volume for stream
maprcli volume create -name streamVolume -path /streamVolume
                
// Create stream (default partitions: 1)
maprcli stream create -path /streamVolume/changelogStream -ischangelog true
                
// Create stream  (default partitions: 3)
maprcli stream create -path /streamVolume/changelogStream -ischangelog true -defaultpartitions 3
// Create Volume for stream
https://10.10.100.17:8443/rest/volume/create?name=streamVolume&path=/streamVolume

// Create stream (default partitions: 1)
https://10.10.100.17:8443/rest/stream/create?path=/streamVolume/changelogStream&ischangelog=true

// Create stream (default partitions: 3)
https://10.10.100.17:8443/rest/stream/create?path=/streamVolume/changelogStream&ischangelog=true&defaultpartitions=3

Create Topic

An MapR Event Store For Apache Kafka stream topic must be created for the changed data records. This can be accomplished in a variety of ways:
  • Use the maprcli table changelog add command. This command establishes a changelog relationsip between the source table and the destination stream topic.
  • Use the maprcli stream topic create command. I
  • Use the REST equivalent of the above maprcli commands.
  • Use the Control System.
IMPORTANT Once a changelog relationship is established between the source table and the destination stream topic, the number of topic partitions is locked. (The maprcli table changelog add command is used to establish the changelog relationship.) The stream topic edit command can not be used to modify the topic's number of partitions.
The following describes when to create a stream topic a specific way.
  • If the changelog stream's default partitions are acceptable for the stream topic (because the topic inherits the stream's default partitions), you can either:
    • Go directly to adding the changelog relationship with the maprcli table changelog add command and create the topic there.
    • Create the topic with the stream topic create command and not specify the -partitions parameter.
  • If you want to change the topic's partitions, create the topic with the stream topic create command and set the -partitions parameter.
  • If you use the Control System, either of the above methods are available.

The following code examples show how to create a stream topic and change the default partition to five (5). 

// Create topic (default partitions: 5
maprcli stream topic create -path /streamVolume/changelogStream -topic cdcTopic1 -partitions 5
// Create topic (default partitions: 5
https://10.10.100.17:8443/rest/stream/topic/create?path=/streamVolume/changelogStream&topic=cdcTopic1&partitions=5

Add Changelog

A table changelog relationship must be added between the source table and the destination stream topic by using the maprcli table changelog add command or the Control System. By adding a table changelog relationship, you are creating an environment that propagates changed data records from a source table to an MapR Event Store For Apache Kafka stream topic.

  • If you are creating a changelog relationship and the stream topic does not exist, specify the stream path and topic.
  • If you are creating a changelog relationship and the stream topic does exist, specify the stream path and topic AND the -useexistingtopic parameter. The -useexistingtopic parameter can only be used with a changelog stream's newly created topic or a previous changelog stream topic for the same source table.
NOTE Propagation of existing table data is enabled by default. If you do not want to propagate existing source table data, set the -propagateexistingdata parameter to false. The default is true.
NOTE Propagation is enabled as soon as the table changelog relationship is added. If you do not want propagation to begin, set the -pause parameter to true. The change data records are stored in a bucket until you resume the changelog relationship; at this point, the stored change data records are propagated to the stream topic. See table changelog resume for more information.
The following examples show you how to:
  • Create a changelog relationship between the source table and the destination stream topic, where the stream topic does not exist.
  • Create a changelog relationship between the source table and the destination stream topic, where the stream topic does exist.
IMPORTANT The examples show streams in a local cluster. To specify streams in a destination (remote) cluster, the path format is /mapr/<remote-cluster>/path/to/stream:topic. 
maprcli table changelog add -path /tableVolume/cdcTable -changelog /streamVolume/changelogStream:cdcTopic1
                  
maprcli table changelog add -path /tableVolume/cdcTable -changelog /streamVolume/changelogStream:cdcTopic1 -useexistingtopic true
https://10.10.100.17:8443/rest/table/changelog/add?path=/tableVolume/cdcTable&changelog=/streamVolume/changelogStream:cdcTopic1

https://10.10.100.17:8443/rest/table/changelog/add?path=/tableVolume/cdcTable&changelog=/streamVolume/changelogStream:cdcTopic1&useexistingtopic=true

The following example verifies that the table changelog relationship exists:

maprcli table changelog list -path /tableVolume/cdcTable

What's Next: Modifying and Consuming Data

To have CDC changed data records to consume, you need to perform inserts, updates, and deletes on the MapR Database table data. See CRUD operations on documents using mapr dbshell for JSON documents, mapr hbshell for binary data, Java applications for MapR Database JSON, C or Java applications for MapR Database Binary.

An MapR Event Store For Apache Kafka Kafka/OJAI consumer application subscribes to the topic and consumes the change data records. See Consuming CDC Records for more information.