Hi,

I want your opinion/suggestion on what can be done in the below mentioned case.

Is it true that for replication to work in DB2, at least one unique index is required on the tables selected to replicate. I have tables in the database which do not have anu unique index and I cannot create also.

Request your suggestion/alternative solution to resolve this problem.

Regards,
Gulshan

Gulshan,
Yes there needs to be a unique index. How do yo expect DB2 to Update or Delete the proper row without one?

Andy

Can't you just slap an identity column on it?

Does slapping identity columns means addition of a new column in the table that would be of integer type and incerement itself by 1 on addition of each row ? The new coulmn can be the unique index for the table.
If yes, then it would require change in 20 odd database tables and changes in the application code as well

Are there any other alternatives ?

If you use an Identity column in a table that is being replicated, make sure that you use GENERATE BY DEFAULT and not GENERATE ALWAYS. You do not want to have your identity columns change values on the target server.

Replication requires a unique index to determine which row needs changing. There is no other way around it. If you cannot provide some method of uniqueness for a table that Replication is not the answer you are looking for.

Andy

Thanks for all the help. I would use log forwading instead.

just wonder u wanna replica copy ( exact copy of the table in more than one DB ) ?

or u just need to collect data from some tables to another ?

i had wrote ODBC based data collection tool under linux and i can provide you with a copy of it

good luck

Yes, I want to copy the exact data from one database to another. I am using AIX as O/S and CICS as OLTP.