Does Varada accelerate Hive views?

A Hive view is built on top of other tables (or views) and does not hold data itself. Therefore, when you query a Hive view in the Varada catalog, the data retrieval has already taken place and Varada cannot apply any acceleration.

If you frequently query views and want to experience the power of Varada acceleration, as a workaround you can either change the view definition to force the system to query the underlying tables in the Varada catalog, as described in Accelerate Hive Views by Updating the View Definition or rename the catalogs so that all queries use the Hive catalog, as described in Accelerate Hive Views by Renaming the Catalogs.

Accelerate Hive Views by Updating the View Definition

You can enable Varada to accelerate Hive views by changing the view definition to force the system to query the underlying tables in the Varada catalog. To do this, You rename the hive catalog in the view definition to the varada catalog.

For example, if the view definition is:

create view sales as select * from hive.sales.tbl1

change it to:

create view sales as select * from varada.sales.tbl1

Accelerate Hive Views by Renaming the Catalogs

You can enable Varada to accelerate Hive views by using the Varada connector with the Hive catalog. To do this:

  1. Rename the Hive catalog hive.properties to old_hive.properties or any other name of your choice.

  2. Rename the Varada catalog varada.properties to hive.properties.

  3. Restart the cluster.