Table of Contents

Search

  1. Abstract
  2. Supported Versions
  3. Performance Tuning and Sizing Guidelines for Informatica® Big Data Management 10.2.2

Performance Tuning and Sizing Guidelines for Informatica® Big Data Management 10.2.2

Performance Tuning and Sizing Guidelines for Informatica® Big Data Management 10.2.2

Data Processor Transformation

Data Processor Transformation

Consider the following best practices for mappings that contain a Data Processor transformation:
  • For mappings that run on the Blaze engine, it is required to enable partitioning for Data Processor transformations. To enable partitioning, perform the following steps:
    1. In the Developer tool, open the mapping and select the Data Processor transformation.
    2. In the
      Properties
      view, click the
      Advanced
      tab.
    3. Select
      Enable partitioning for Data Processor transformations
      .
      The following image shows the Advanced tab of a Data Processor transformation:
  • When a mapping with a Data Processor transformation meets all of the following conditions, the Blaze engine processes the entire mapping in a single tasklet:
    • The mapping source file is of a non-splittable input format.
    • The transformation contains multiple output groups.
    The Data Processor transformation might output a higher data volume than the source. For such scenarios, configure the Blaze engine to first stage the data generated by the transformation at each output group.
    The following image shows a mapping with a Data Processor transformation with multiple output groups:
    To stage data at every output group, set the following mapping run-time property in the Developer tool:
    Parameter
    Value
    Blaze.StageOutputGroupDataForInstances
    The name of the Data Processor transformation instance.
    When the Blaze engine is configured to first stage the data, it performs the following tasks:
    • Re-partitions the data.
    • Processes the staged data.
    • Creates the correct number of tasklets based on the staged data volume.

0 COMMENTS

We’d like to hear from you!