Oracle 8i to MS SQL Server replication

Question

Im new to Oracle, but a MS SQL Server DBA. Im helping in the architecture of a data warehouse that pulls from some disparate sources, including Oracle 8i, into a SQL Server 2000 box. We could use batch processing to pull updated data from the Oracle database to populate our data warehouse, but we were hoping to get real-time transactional processing via some replication scenario. My terminology may be wrong from the Oracle perspective, but I think the transactional replication that SQL Server supports is called synchronous replication in Oracle. My question is, does anybody know how to do synchronous replication from an Oracle server to a SQL Server? Or perhaps which Oracle product supports it? Thanks!
null

Answer 1

===========================================================
I'm working on the similar requirements, exactly my source DB is DB2, and the target DB is Oracle. I have two year experience with Oracle, but I know nothing about DB2. There a about 50 tables in source DB. Transactions on source DB is not too heavy(less than 10/Second), and the mainfram are all powerful.
I've though about Oracle Gateway, but I'd rather save it as the backup solution. For one thing, neither co-workers nor I have experience with it. Second, I'm afraid the Gateway performance would not be satisfactory. Furthermore, words has it that we should pay for the Gateway. So My ideas to real-time replication of the two DB is as following:
(1) use trigger in DB2 to reponse to immediate transactions in source DB;
(2) use Embedded C/SQL to extract those transactions into flat files;
(3) FTP files to the target server;
(4) On target server, SHELL scripts read the files and call Pro*C to apply the transactions into Oracle DB.
but, some problems really get me.
A: in step(1) and (2), How to set EC to do its work in reponse to trigger? I know that one can call host script in Oracle trigger, but I have no idea about DB2 trigger. Maybe, I can create a real-time monitoring EC process to address it. what's your opinion?
B: I should maintain all transaction metadata by myself. e.g. INSERT, UPDATE old/new data, DELETE info. It is originally covered by the database, and it isn't easy to handle it squarely.
C: How to address some exception? like database rollback, the process recovery in case of any interruptions in the above steps.
Could you share me some idea, Thank you!
eilison
[email protected]