Table Partitioning

Hi,

One of my ETL sessions is taking too long to execute. The session actually creates indexes on the table and due to excessive data volume (somewhere around 90 million rows) of the table, it takes too long to execute. I've been advised to partition this table so that index creation is done only on the incremental data that is being loaded thereby avoiding the redundant task of dropping and creating indexes on already loaded data. However, i am confused about deciding the partitioning criteria for the table. Kindly educated me about the core concept of partitioning and how should i