Table of Contents

Search

  1. Preface
  2. Mappings
  3. Parameters
  4. CLAIRE recommendations
  5. Data catalog discovery
  6. Visio templates

Mappings

Mappings

Rules and guidelines for mappings on GPU-enabled clusters

Rules and guidelines for mappings on GPU-enabled clusters

Use the following rules and guidelines when you configure a mapping that runs on a GPU-enabled elastic cluster:
  • The size of a partition file must be smaller than the GPU memory size. To check the GPU memory size, refer to the AWS documentation for the selected worker instance type.
  • The mapping cannot read from an Amazon Redshift source or a source based on an intelligent structure model.
  • The mapping cannot write to a CSV file.
  • If the mapping includes NaN values, the output is unpredictable.
  • The mapping cannot process timestamp data types from a Parquet source.
  • If you need to process decimal data from a CSV file, read the data as a string and then convert it to a float.
  • When the mapping uses an Expression transformation, you can use only scientific functions and the following numeric functions:
    • ABS
    • CEIL
    • FLOOR
    • LN
    • LOG
    • POWER
    • SQRT
To learn how to configure a GPU-enabled cluster, refer to
Data Integration Elastic Administration
in the Administrator help.