source
Lakehouse Source Connector
pulsar lakehouse connector
Authored by
hangc0276,zymap,Huanli-Meng,horizonzy
Support type
streamnative
License
Apache License 2.0

The Lakehouse source connector (currently only including the Delta Lake source connector) fetches the Lakehouse table's changelog and saves changelogs into a Pulsar topic.

How to get

This section describes how to build the Lakehouse source connector.

You can get the Lakehouse source connector using one of the following methods:

To build the Lakehouse source connector from the source code, follow these steps.◊

  1. Clone the source code to your machine.

    git clone https://github.com/streamnative/pulsar-io-lakehouse.git
    
  2. Build the connector in the pulsar-io-lakehouse directory.

    • Build the NAR package for your local file system.

      mvn clean install -DskipTests
      
    • Build the NAR package for your cloud storage (Including AWS, GCS, and Azure-related package dependency).

      mvn clean install -P cloud -DskipTests
      

    After the connector is successfully built, a NAR package is generated under the target directory.

    ls target
    pulsar-io-lakehouse-2.11.2.13.nar
    

How to configure

Before using the Lakehouse source connector, you need to configure it. This table lists the properties and the descriptions.

NameTypeRequiredDefaultDescription
typeStringtrueN/AThe type of the Lakehouse source connector. Available values: delta.
checkpointIntervalintfalse30The checkpoint interval (in units of seconds). By default, it is set to 30s.
queueSizeintfalse10_000The buffer queue size of the Lakehouse source connector. The buffer queue is used for store records before they are sent to Pulsar topics. By default, it is set to 10_000.
fetchHistoryDataboolfalsefalseConfigure whether to fetch the history data of the table. By default, it is set to false.
startSnapshotVersionlongfalse-1The Delta snapshot version to start capturing data change. Available values: [-1: LATEST, -2: EARLIEST]. The startSnapshotVersion and startTimestamp are mutually exclusive.
startTimestamplongfalseN/AThe Delta snapshot timestamp (in units of seconds) to start capturing data change. The startSnapshotVersion and startTimestamp are mutually exclusive.
tablePathStringtrueN/AThe path of the Delta table.
parquetParseThreadsintfalseRuntime.getRuntime().availableProcessors()The parallelism of paring Delta Parquet files. By default, it is set to Runtime.getRuntime().availableProcessors().
maxReadBytesSizeOneRoundlongfalseTotal memory * 0.2The maximum read bytes size from Parquet files in one fetch round. By default, it is set to 20% of the heap memory.
maxReadRowCountOneRoundintfalse100_000The maximum read number of rows processed in one round. By default, it is set to 1_000_000.

Note

The Lakehouse source connector uses the Hadoop file system to read and write data to and from cloud objects, such as AWS, GCS, and Azure. If you want to configure Hadoop related properties, you should use the prefix hadoop..

Examples

You can create a configuration file (JSON or YAML) to set the properties if you use Pulsar Function Worker to run connectors in a cluster.

  • The Delta table that is stored in the file system

    {
        "tenant":"public",
        "namespace":"default",
        "name":"delta_source",
        "parallelism":1,
        "topicName": "delta_source",
        "processingGuarantees":"ATLEAST_ONCE",
        "archive": "connectors/pulsar-io-lakehouse-2.11.2.13.nar",
        "configs":{
            "type":"delta",
            "checkpointInterval": 180,
            "queueSize": 10000,
            "fatchHistoryData": false,
            "startSnapshotVersion": -1,
            "tablePath": "file:///tmp/data/delta-source",
            "parquetParseThreads": 3,
            "maxReadBytesSizeOneRound": 134217728,
            "maxReadRowCountOneRound": 100000
        }
    }
    
  • The Delta table that is stored in cloud storage (AWS S3, GCS, or Azure)

    {
        "tenant":"public",
        "namespace":"default",
        "name":"delta_source",
        "parallelism":1,
        "topicName": "delta_source",
        "processingGuarantees":"ATLEAST_ONCE",
        "archive": "connectors/pulsar-io-lakehouse-2.11.2.13-cloud.nar",
        "configs":{
            "type":"delta",
            "checkpointInterval": 180,
            "queueSize": 10000,
            "fatchHistoryData": false,
            "startSnapshotVersion": -1,
            "tablePath": "s3a://test-dev-us-west-2/lakehouse/delta_source",
            "hadoop.fs.s3a.aws.credentials.provider": "com.amazonaws.auth.DefaultAWSCredentialsProviderChain",
            "parquetParseThreads": 3,
            "maxReadBytesSizeOneRound": 134217728,
            "maxReadRowCountOneRound": 100000
        }
    }
    

Data format types

Currently, The Lakehouse source connector only supports reading Delta table changelogs, which adopt a parquet storage format.

How to use

You can use the Lakehouse source connector with Function Worker. You can use the Lakehouse source connector as a non built-in connector or a built-in connector.

If you already have a Pulsar cluster, you can use the Lakehouse source connector as a non built-in connector directly.

This example shows how to create a Lakehouse source connector on a Pulsar cluster using the pulsar-admin sources create command.

PULSAR_HOME/bin/pulsar-admin sources create \
--source-config-file <lakehouse-source-config.yaml>

Demos

This table lists demos that show how to run the Delta Lake, Hudi, and Iceberg source connectors with other external systems.

Currently, only the demo on the Delta Lake source connector is available.

ConnectorLink
Delta LakeFor details, see the Delta Lake demo.
Hudi
Iceberg