A merge table supported for import processing must satisfy the following conditions:

• It must be a general table, not a time-ordered information table.
• It must be a non-FIX table. If it has the FIX attribute, the null value storage option is not specified with DELETE. This condition is not applicable if a merge table is not created from n tables.
• Import processing must use an SQL process. You cannot create a merge table for import processing if data has been edited by an import information editing UOC routine.
• The mapping keys must be the same (applicable to a merge table from n tables). There must also be a unique index of all or part of the mapping key, and there must be no unique index for non-mapping key columns.
• PURGE TABLE cannot be executed on the extracted table (applicable to a merge table from n tables).
• There can be no updating of the mapping key (applicable to a merge table from n tables).

A merge table is created according to the following rules:

• If UPDATE data is not found, it is stored as INSERT data.
• If INSERT data results in a key value duplication, the data is stored as UPDATE data.
• For data subject to DELETE, either the corresponding row is deleted or the null value is placed in the corresponding column, depending on an import definition option.

If you execute UPDATE specifying an element number for a table containing a repetition column, only the updated element data is sent as the update information, not the entire element data in the corresponding column. In this case, Datareplicator compensates the table with the null value so that there will be enough data items for INSERT during merge table import processing. The following table shows this compensation method:

Note

Element # indicates the element number. Maximum element # and Minimum element # indicate the maximum and minimum element number contained in the update information for the corresponding repetition column.

• When concatenation operation data is imported into a merge table and there is no corresponding row for UPDATE (SQLCODE=100), only the concatenation operation data is imported by INSERT.
• Do not import into a merge table BLOB-type or BINARY-type data to which backward deletion updating is applied. Inconsistency might result between the source and target databases.