When to Use Redshift RA3 vs Spectrum

Both have architectures where the compute and storage layers are separated, raising the question of which to use for which use case.

• Examples

  • RA3 patterns:
    • When offloading data to S3 to save storage, use RA3
    • When performing ETL within Redshift to process data
    • When you don’t want to worry about Spectrum scan performance such as partitioning on S3
  • Spectrum:

    • When accessing data on S3 from other AWS services. If data is not currently in DWH, use Spectrum. (In this case, deciding between Athena is also a consideration…)

    • When processing data with EMR and referencing data with Redshift

    • When combining with data within Redshift