Cloudera Docs

Migrating Impala workloads

Cloudera Data Warehouse (CDW) Private Cloud offers the latest, updated, and managed elastic Impala experience. Migrating the Impala workloads to CDW enables you to leverage auto-scaling, data sharing, data and result caching, and many other powerful capabilities of the Data Lakehouse architecture.

If you are on CDH or HDP platforms, then you must first upgrade your clusters to CDP Private Cloud Base before you can install CDP Private Cloud Data Services and migrate your Impala compute workloads to CDW. Compute workloads are SQL programs that get generated from either a shell script, a JDBC/ODBC client, thick SQL clients, or Hue.

Moving certain Impala workloads from CDP to CDW has the following advantages:
  • Ability to scale out the Impala compute layer independently from the storage layer
  • Provides better isolation between workloads to prevent interference and maintain SLAs
  • Simplifies the management of Impala clusters, by not having you to focus on admission control and tuning configuration parameters
  • Automatic installation, setup, start-up, scale-up, and scale-down of workloads running on a Kubernetes backend
  • Ability to run different versions of Impala for different types of workloads (For example, operational reporting can remain locked on a given version, while data science users can always upgrade their Virtual Warehouses to the latest versions)