Some tables stay the same size after they are created, while others change frequently. A table should use a map that has the number of AMPs appropriate to the table size, which leads to an even distribution of the table rows among the AMPs in the map. Consequently, it is a good idea to periodically reassess a table to ensure that it uses a map appropriate for the current size and for expected growth.
Here are some scenarios in which a DBA might want to use maps to move table data from one set of AMPs to another.
System Expansion
After you add AMPs or nodes to a system, if MAPS architecture is enabled, the Reconfiguration utility (Reconfig) creates a new map that reflects the expanded system. A DBA can then choose to move table rows to the new AMPs using either Reconfig or MAPS. There are several advantages to using MAPS instead of Reconfig:
- Reconfig requires more planned downtime.
- Reconfig requires you to move all tables in the system immediately to evenly distribute rows between AMPs. However, if you use MAPS, you can move tables into the new configuration gradually while the system is online. Some tables can be in the old map and some tables can be in the new, expanded map.
The Configuration and Reconfiguration utilities must be run under the supervision of Teradata support center personnel.
You can also use the MAPS Advisor procedures to model the moves before a system expansion. You can redo the analysis as many times as needed.
Upgrade
When you upgrade your system to Release 16.10 or later and you enable Teradata MAPS architecture, you can look for small tables that would benefit from using a sparse map.