Partitioning: Why Scanning Less Data Matters

Reading the sketch

1. Review how much data is scanned

   Check whether the query reads data beyond what it needs for the result.

2. Organize data into partitions

   Partitioning separates data into groups that the query engine may be able to read independently.

3. Skip irrelevant partitions

   The benefit depends on whether the query can avoid reading partitions it does not need. A query that scans most partitions may gain little.

4. Check the actual result

   Compare the scanning work and runtime after the change. Cost savings depend on the platform and how it charges for the workload.