Fragmenting by ROUND ROBIN

In a round-robin distribution scheme, specify at least two dbspaces where you want the fragments to be placed, or specify at least two fragment names in one or more dbspaces. As records are inserted into the table, they are placed in the first available fragment. The database server balances the load among the specified fragments as you insert records and distributes the rows in such a way that the fragments always maintain approximately the same number of rows (±100). A newly-added round-robin fragment will be favored exclusively by inserts and loads until it no longer has the fewest rows among the table’s fragments.

In this distribution scheme, the database server must scan all fragments when it searches for a row.

Important:

The FRAGMENT BY ROUND ROBIN clause overrides the automatic location and fragmentation of tables, which is enabled when the AUTOLOCATE configuration parameter or the AUTOLOCATE session environment option is set to a positive integer.

When automatic location and fragmentation is enabled, the database server automatically determines
  • the table extent sizes,
  • the dbspaces where the fragments are stored,
  • and a ROUND ROBIN distributed-storage strategy for the new table.

Round-robin fragmentation with large object data types

For simple large objects in tables that contain BYTE or TEXT columns and that are fragmented by round-robin, you can reserve space for inserting BYTE and TEXT data by setting the PN_STAGEBLOB_THRESHOLD configuration parameter. For information about how the database server stages simple large objects in round-robin fragments, see the description of PN_STAGEBLOB_THRESHOLD in your HCL OneDB™ Administrator's Reference.

For smart large objects in tables that contain BLOB or CLOB columns, you can use the PUT clause to specify round-robin fragmentation in a list of sbspaces. When you include the PUT clause in the CREATE TABLE statement (or in the CREATE TEMP TABLE statement or in the ALTER TABLE statement), you have the options to include or not include a FRAGMENT BY clause defining distributed storage for other columns in the same table. The PUT clause applies a round-robin storage distribution strategy only to smart large object columns for which it specifies more than one sbspace. For more information and examples, see the PUT Clause.