Connect CDC SQData

Replication to AVRO containers in either plain file or HDFS (Hadoop file system) can be accomplished with minimal configuration and provide excellent performance.

Prerequisites

1.AVRO container targets are restricted to a single worker and each target must have its own individual file.

2.As a consequence the target url must be a generic url and the substitution parameter for the generic url must be unique for each target.

3.By default, the qualified name of the source is used as substitution parameter. That can be overridden using the MAPPINGS section, either static or dynamic.

Example 1

AVRO Container formatted file or HDFS (Hadoop file system) where every source object (i.e. table name) will be written to a unique file using AVRO CONTAINER formatting and default file Rotation.

REPLICATE

  DB2 cdc://<host_name>:<sqdaemon_port>/<publisher_name>/<subscription_name>

  TO AVRO CONTAINER [file | hdfs]:///*

;

 

Example 2

AVRO Container formatted file or HDFS (Hadoop file system) where every source object name (i.e. table name) is written to a specified file name with a common prefix and suffix and specific file rotation parameters in specified in an OPTIONS statement.

REPLICATE

  DB2 cdc://<host_name>:<sqdaemon_port>/<publisher_name>/<subscription_name>

  TO AVRO CONTAINER [file | hdfs]:///<prefix>_*_<suffix>

;

 

OPTIONS

ROTATE SIZE 100M,

ROTATE DELAY 30

;

 

Note: AVRO Container targets use a file rotation method controlled by a size and or a delay. For hadoop (HDFS), if an OPTIONS statement does not define a size and delay, by default a delay of 1 hour is applied. A delay of one hour means that from the time a record is first written to a target file, the file will be rotated after one hour has passed, if the file has not been rotated for other reason.