MEM_SIZE varies significantly for string columns where a value is dominant

Details

Detail name Value
Changelog Number 16931
Type Bug
Status Resolved
Affected Versions
Fix Versions Exasol 8.19.0
Resolution Date 2023-06-15

Background

Exasol uses different compression algorithms depending on the data distribution.

Description

If the following conditions are met, the compressed size (MEM_SIZE) of string columns may have significant differences, even though the data in the columns are the same:

  • The column is of type VARCHAR or CHAR UTF8
  • A value occurs significantly more often than all the other values in the column

Workaround

Recompress the table/columns using the RECOMPRESS statement. For example:

RECOMPRESS TABLE TAB1 (COL1) ENFORCE;

Fix

In these scenarios, there is no more significant difference in the compressed size of a column.