Another scenario where partition is good. We have a data warehouse
database with many fact tables loaded daily from more than 100
companies worldwide. One of the tables is an "item" table where the PK
is COMPANY_CODE and ITEM_CODE. We partitioned the table by company and
after we load, check and tranform the data sent by a specific company
in a temporary table (with the same structure) we just exchange this
table with the specific company partition (ALTER TABLE ... EXCHANGE
PARTITION). It takes just 1/100 second, minimum availability
interruption, minimal lock, etc.
Received on Fri Apr 08 2005 - 11:44:22 CDT
