Buffered-insert imports

In a partitioned database environment, the import utility can be enabled to use buffered inserts. This reduces the messaging that occurs when data is imported, resulting in better performance.

The buffered inserts option should only be enabled if you are not concerned about error reporting, since details about a failed buffered insert are not returned.

When buffered inserts are used, import sets a default WARNINGCOUNT value to 1. As a result, the operation will fail if any rows are rejected. If a record is rejected, the utility will roll back the current transaction. The number of committed records can be used to determine which records were successfully inserted into the database. The number of committed records can be non zero only if the COMMITCOUNT option was specified.

If a different WARNINGCOUNT value is explicitly specified on the import command, and some rows are rejected, the row summary output by the utility can be incorrect. This is due to a combination of the asynchronous error reporting used with buffered inserts and the fact that an error detected during the insertion of a group of rows causes all the rows of that group to be backed out. Since the utility would not reliably report which input records were rejected, it would be difficult to determine which records were committed and which records need to be re-inserted into the database.

   db2 connect to your_database
   db2 bind db2uimpm.bnd insert buf

Buffered inserts feature cannot be used in conjunction with import operations in the INSERT_UPDATE mode. The bind file db2uImpInsUpdate.bnd enforces this restriction. This file should never be bound with the INSERT BUF option. This causes the import operations in the INSERT_UPDATE mode to fail. Import operations in the INSERT, REPLACE, or REPLACE_CREATE modes are not affected by the binding of the new file.