April 21, 2014 \ Ananth TM Teradata TPUMP stage · TPump is a highly parallel utility designed to continuously move data from data sources into Teradata. · TPump is typically used for loading a small quantity of records in relation to the size of your target table. · TPump works at the row level, whereas MultiLoad and FastLoad update whole blocks of data. TPump allowing us to load data into tables with referential integrity which MultiLoad doesn’t. · TPump only needs to take row-level locks; in other words, TPump only places a lock upon the row it is modifying. In contrast, MultiLoad places locks on the entire table. If you need multiple processes updating a table simultaneously, TPump may be the better solution. · TPump also uses fewer system resources, so you can run it concurrently with user queries without impacting system performance. · TPump lets you control the rate at which updates are applied. You can dynamically “throttle down” when the system is busy and throttle up when activity is lighter. · TPump does not take a utility slot. · TPump is designed for “trickle feed” taking individual row-level locks. If you use TPUMP, you need to make sure that you follow normal TPUMP standards (Set the KEY statements equal to the PI and turn SERIALIZE on). · Tpump is typically for processes that are constantly retrieving and processing data, like from a message queue. · Using TPump offers a controllable transition to updating that is closer to real time. Best fit in “Active”, “real-time” or “closed-loop” data warehousing.