Hi,
I need to import a few tables from one database to the other. But I would ideally like to use the insert_update option since almost 90% of the rows are already in the tables. So i want them to be just updated if they are changd. But my problem is that most of these tables dont have either a primary key or a unique key. There is no way for DB2 to knowwhich rows to check for updates.

1) Is it a necessitiy to have a primary key?
2) Is there any other way to replicate the databases?

HADR is not a solution since we need both databases.
i guess even db2 replication needs primary keys.
My main objective is to replicate one DB to another which will be used during day time. So in the night I need to replicate.

DB2 v8.2 on AIX 5.3

Is there a way for _you_ to know which rows to update? If yes, use this knowledge to create a primary key. If not, add an artificial primary key column.

No I wont know which rows will be updated.