Table of Contents

Search

  1. Preface
  2. Introduction to Databricks Delta Connector
  3. Connections for Databricks Delta
  4. Mappings for Databricks Delta
  5. Migrating a mapping
  6. Databricks Delta SQL ELT optimization
  7. Data type reference

Databricks Delta Connector

Databricks Delta Connector

Single commit for SQL ELT optimization

Single commit for
SQL ELT optimization

When you enable full
SQL ELT optimization
for a mapping to write to multiple Databricks Delta targets, you can configure the mapping to commit the configured operations for all the targets within a connection group together.
You can use single commit to combine the metadata from all the targets and send the metadata for processing in a single execution call. When you use single commit, the Secure Agent separates the targets into connection groups based on equivalent connection attributes and commits the operations together for each connection group. This optimizes the performance of the write operation.
When you run a mapping with multiple targets, the Databricks Delta connections used for these multiple target transformations that have the same connection attribute values are grouped together to form connection groups. As all the targets in a connection group have the same connection attributes, only a single connection is established for each connection group which represents that particular connection group. The transactions on each connection group runs on a single Databricks cluster.
If the Secure Agent fails to write to any of the targets, the task execution stops and the completed transactions for the targets that belong to the same connection group are not rolled back.
To enable single commit to write to multiple targets, set the
EnableSingleCommit=Yes
custom property in the
Advanced Session Properties
section on the
Runtime Options
tab of the mapping task.
When you run a mapping with single commit enabled, you can view the row statistics details in the session logs.
Single commit is applicable only when you run a mapping on Databricks cluster.

0 COMMENTS

We’d like to hear from you!