Flexible data clustering refers to sorting table data along naturally occuring dimensions, such as name, date, or location. Data clustering optimizes table structure to significantly improve query performance, especially on very large tables. A well-clustered table increases the effectivity of the metadata collected by focusing on a specific and limited range of rows, called chunks.
The following are some scenarios ideal for data clustering:
Queries containg a
WHEREpredicate written as
column COMPARISON value, such as
date_column > '2019-01-01'or
id = 107when the columns referenced are clustering keys.
In such a case SQream reads the portion of data that contain values matching these predicates only.
Two clustered tables joined by their respective clustering keys.
In such a case SQream uses metadata to more easily identify matching chunks.