Skip to main content

Syncing to Unity Catalog

This document walks through the steps to register an Apache XTable™ (Incubating) synced Delta table in Unity Catalog on Databricks and open-source Unity Catalog.

Pre-requisites (for Databricks Unity Catalog)​

  1. Source table(s) (Hudi/Iceberg) already written to external storage locations like S3/GCS/ADLS. If you don't have a source table written in S3/GCS/ADLS, you can follow the steps in this tutorial to set it up.
  2. Setup connection to external storage locations from Databricks.
    • Follow the steps outlined here for Amazon S3
    • Follow the steps outlined here for Google Cloud Storage
    • Follow the steps outlined here for Azure Data Lake Storage Gen2 and Blob Storage.
  3. Create a Unity Catalog metastore in Databricks as outlined here.
  4. Create an external location in Databricks as outlined here.
  5. Clone the Apache XTable™ (Incubating) repository and create the xtable-utilities-0.2.0-SNAPSHOT-bundled.jar by following the steps on the Installation page

Pre-requisites (for open-source Unity Catalog)​

  1. Source table(s) (Hudi/Iceberg) already written to external storage locations like S3/GCS/ADLS or local. In this guide, we will use the local file system. But for S3/GCS/ADLS, you must add additional properties related to the respective cloud object storage system you're working with as mentioned here
  2. Clone the Unity Catalog repository from here and build the project by following the steps outlined here

Steps​

Running sync​

Create my_config.yaml in the cloned Apache XTable™ (Incubating) directory.

yaml
sourceFormat: HUDI|ICEBERG # choose only one
targetFormats:
  - DELTA
datasets:
  -
    tableBasePath: s3://path/to/source/data
    tableName: table_name
    partitionSpec: partitionpath:VALUE # you only need to specify partitionSpec for HUDI sourceFormat
Note:
  1. Replace s3://path/to/source/data to gs://path/to/source/data if you have your source table in GCS and abfss://<container-name>@<storage-account-name>.dfs.core.windows.net/<path-to-data> if you have your source table in ADLS.
  2. And replace with appropriate values for sourceFormat, and tableName fields.

From your terminal under the cloned Apache XTable™ (Incubating) directory, run the sync process using the below command.

shell
java -jar xtable-utilities/target/xtable-utilities-0.2.0-SNAPSHOT-bundled.jar --datasetConfig my_config.yaml
Note:

At this point, if you check your bucket path, you will be able to see _delta_log directory with 00000000000000000000.json which contains the logs that helps query engines to interpret the source table as a Delta table.

Register the target table in Databricks Unity Catalog​

(After making sure you complete the pre-requisites mentioned for Databricks Unity Catalog above) In your Databricks workspace, under SQL editor, run the following queries.

SQL
CREATE CATALOG xtable;

CREATE SCHEMA xtable.synced_delta_schema;

CREATE TABLE xtable.synced_delta_schema.<table_name>
USING DELTA
LOCATION 's3://path/to/source/data';
Note:

Replace s3://path/to/source/data to gs://path/to/source/data if you have your source table in GCS and abfss://<container-name>@<storage-account-name>.dfs.core.windows.net/<path-to-data> if you have your source table in ADLS.

Validating the results​

You can now see the created delta table in Unity Catalog under Catalog as <table_name> under synced_delta_schema and also query the table in the SQL editor:

SELECT * FROM xtable.synced_delta_schema.<table_name>;

Register the target table in open-source Unity Catalog using the CLI​

(After making sure you complete the pre-requisites mentioned for open-source Unity Catalog above) In your terminal start the UC server by following the steps outlined here

In a different terminal, run the following commands to register the target table in Unity Catalog.

shell
bin/uc table create --full_name unity.default.people --columns "id INT, name STRING, age INT, city STRING, create_ts STRING" --storage_location /tmp/delta-dataset/people

Validating the results​

You can now read the table registered in Unity Catalog using the below command.

shell
bin/uc table read --full_name unity.default.people

Conclusion​

In this guide we saw how to,

  1. sync a source table to create metadata for the desired target table formats using Apache XTable™ (Incubating)
  2. catalog the data in Delta format in Unity Catalog on Databricks, and also open-source Unity Catalog
  3. query the Delta table using Databricks SQL editor, and open-source Unity Catalog CLI.