I constantly have discussion with customers regarding GG. Some questions that was raised in a recent meet and what I think. Please help clarify.
1. Can GoldenGate (for Oracle) use Online Redo log without having to use Archive log ?
-- Oracle Databases must be in ARCHIVELOG
mode so that Extract can process the log files. So, I m not sure on using Online Redo log without the need of Archive log ?
2. Can Replicat and Data Pump access an ftp location ?
-- I believe this shouldn't be a problem as long as both process can access the ftp location.
3. Size of the trail file and replication speed. What if the Extract/Pump Trail file is huge and takes time to fill up. Will this cause a lag on Replication ?
-- Shouldn't as I don't think replicat would wait for the file to fill up before processing ? your thoughts ?
4. Initial Load Oracle to SQL Server takes 8 hrs for 200 million rows on a single table .
--- I suggested to use BATCHSQL param. Any other suggestions for Initial Load. Not sure if Unique Index will help on the Target for Inserts ?
5. Why we need these GRANTS for GG for Oracle, explain.
GRANT EXECUTE on DBMS_FLASHBACK to <gguser>; Not sure why.
GRANT ALTER ANY TABLE to <gguser>; -- To Perform administrative changes, such as enabling logging.
GRANT FLASHBACK ANY TABLE to<gguser>; -- Explained here but not clear to me. https://docs.oracle.com/en/middleware/goldengate/core/19.1/oracle-db/preparing-database-oracle-goldengate.html#GUID-AE2D…
6. Pulling data from Multiple Data Sources (all Oracle). Lets say they have 160 Oracle Databases they want to replicate into a single Target(Oracle or SQL Server). How to implement.
-- One GG install would be sufficient. Every Database will have their own set of GG user. There would be 160 extract and 160 replicat Process running. Correct me if I m wrong.
Thank you
Gururaj