What are the steps to add a new fragment to a table?
Fragmentation of a table is performed in order to improve performance of individual queries, to reduce contention between transactions, and to improve data availability
The steps to fragment a table are conditioned by the goal. Decide on the disk layout and strategy before starting:
Try to distribute all of the rows of the table evenly over different disks. Overall query-completion time is reduced when the database server does not need to wait for data retrieval from a table fragment that has more rows than other fragments.
Reduce contention for data in tables that multiple queries and OLTP applications use. Fragmentation can reduce contention when many simultaneous queries against a table perform index scans to return a few rows.
Increased data availability
Distribute table and index fragments across different disks or devices, to improve the availability of data during disk or device failures. The database server continues to allow access to fragments stored on disks or devices that remain operational.
Increased granularity for backup/Restore
The need to limit data unavailability in the event of a disk failure should influence which tables are grouped together in a particular dbspace. Also, the desire to minimize the impact of cold restores might influence the dbspace used to store critical data.
Add a new dbspace to a round robin distribution scheme
Add a new named fragment to a round robin distribution scheme
Add an expression-based fragment
Position a new fragment Before or After an existing fragment
Using a remainder fragment