Compression and storage enhancements
Db2 11.5.7 provides users with the option to improve trickle-feed data insert performance and compression for column-organized tables.
Attention: This mod pack release is currently available for
the following Db2 products:
The following table displays a list of compression and storage enhancements in Db2
11.5.7:
- Db2 on-premises
- The single container deployments of Db2 Warehouse and IBM Integrated Analytics System (IIAS).
- The container micro-service deployment of Db2 on Red Hat OpenShift.
Enhancement | Description |
---|---|
Improved performance when inserting data into column-organized tables, using trickle-feed inserts | With the release of Db2 11.5.7, trickle-feed data insert processing into column organized
tables is improved. A trickle-feed process uses SQL statements to insert a small number of new rows
in one operation. These operations now provide improved processing time and reduced memory, storage,
and log space consumption. Further, inserting data in this way now enables users to apply advanced
compression techniques, such as page-based string compression, to produce better compression results
than in prior releases. In addition, the amount of storage required is reduced when column-organized tables are populated with a small number of rows using insert statements. Data that is inserted using a trickle-feed is usually inserted in an uncompressed state by using the new efficient insert group data page format. This format uses multiple columns that are grouped onto data pages for column-organized tables. In contrast, bulk data insertion statements (SQL statements that insert a large number of new rows) are usually inserted in the traditional fully-compressed column-based data page format. When enough data is inserted in the efficient insert group format to reach an internal threshold, that data is automatically converted into the standard column-based format. The latest compression techniques, including page-based string compression, are applied. For more information, see DB2_COL_INSERT_GROUPS. |