Handling replication for Large & Small volume transactions
765115Nov 3 2010 — edited Mar 25 2011
I want to setup GG to replicate data between Source & Target (both Oracle) databases.
The Source database can have 2 types of transactions at any time -
1) Small size transactions which might involve 4-5 records and which occurs frequently
2) Big volume transaction which might involve 10 million records and which occurs very rarely
I'm planning to use the Extract-Replicat process in GG to implement this replication. I'm concerned that the big volume replication, would choke the subsequent small replications, because, the small replications will start happening only after the big volume replication completes fully.
I'm thinking about the following 2 options. Can someone please advise on these options ?
1) Is there an option to turn on GG in parallel mode to save time ? In parallel mode, will referential integrity be maintained ?
2) Is there an option to replicate the big volume transaction alone in a different way- by not using extract-replicat (because I think this is the slowest for big transactions) and use GG - SQL*Loader combination to load data. I found this approach mentioned in the Administration Guide for replicating initial data load. However I would still prefer to replicate the small transactions via extract-replicat. Given this, is there a way to implement two different replication strategies at the same target.