Transformation without the High-speed Structure Change process
Assume, for example, that you have several non-large partitioned table spaces. You would like to improve your space management by taking advantage of range-partitioned UTSs. You can increase the data set size and convert these table spaces to large partitioned table spaces or range-partitioned UTS table spaces. These changes require that you drop and re-create objects, resulting in an outage.
Without the High-speed Structure Change process, the following steps are required to perform these transformations:
- Start your table space in read-only access.
- Unload the data.
- Drop the old table space.
- Create the new table space and indexes with the new definitions.
- Stop your table space.
- Load the data into the newly defined table space and rebuild all the indexes.
- Perform RUNSTATS.
- Re-create any auxiliary objects that were lost due to dropping the table space, such as views or triggers.
- Rebind the programs that access this table.
Start the table space in read-write access.The data is unavailable for update during Step 1 through Step 4. In Step 5 through Step 9, the data is unavailable for reading and updating. Additionally, this process requires a large amount of CPU time. The UNLOAD and LOAD must process each row in the table space and rebuild each index.
The following figure provides a hypothetical timeline that refers back to these steps for a conversion that requires you to drop and re-create objects:
Related topic