Skip to main content

The paper highlights that Presto does not have its own storage layer. Instead, it uses a Connector API to decouple computation from storage. This allows Presto to query data "in situ" (where it sits).

More verbose but sometimes easier to optimize and debug.

For large tables, bucketing on the join key helps. Presto can perform when both tables are bucketed by the join key with matching bucket counts.

^new^ Full: Presto

The paper highlights that Presto does not have its own storage layer. Instead, it uses a Connector API to decouple computation from storage. This allows Presto to query data "in situ" (where it sits).

More verbose but sometimes easier to optimize and debug. presto full

For large tables, bucketing on the join key helps. Presto can perform when both tables are bucketed by the join key with matching bucket counts. The paper highlights that Presto does not have