Recommended Number of Rows for a Single Batch

For an initial load, you can use 30,000 rows for a large batch. For ongoing loads, you can use 20,000 rows for a large batch. Do not exceed 100,000 rows in a large batch.

Furthermore, for Microsoft SQL Server and Oracle Database environments, limit the number of records in the EIM tables to those that are being processed. For example, if you have determined that the optimal batch size for your implementation is 19,000 rows per batch and you are going to be running eight parallel Siebel EIM processes, then you must have 152,000 rows in the EIM table. Under no circumstances can you have more than 250,000 rows in any single EIM table because this reduces performance. The restrictions mentioned in the preceding example do not apply to IBM DB2 environments. As long as an index is being used to access the EIM tables, the numbers of rows in the EIM tables does not matter in DB2 environments.

Caution: For all supported RDBMS platforms, if indexes are added to EIM tables with any column other than IF_ROW_BATCH_NUM in the first position, then parallel Siebel EIM operations performed on that table will likely fail with index contention issues.

Note: The number of rows that you can load in a single batch can vary depending on your physical computer setup and on which table is being loaded. To reduce demands on resources and improve performance, generally try to vary batch sizes to determine the optimal size for each entity to be processed. In some cases, a smaller batch size can improve performance. But for simpler tables such as S_ASSET, you might find that loads perform better at higher batch sizes than for more complex tables such as S_CONTACT.