Thanks for the replies.
I see there is an evolution of GG to better integrate with the Oracle environment.
There seems to be a new option called:
DBLogReader, DBLogReaderBufSize size, where size can be up to 4MB, it seems that this API was implemented in 10.2.0.5/126.96.36.199 to improve the performance capture process. From what I've read this option uses the database server to access the online redo / archiver logs. Then is not necesary specify ASM instance crediantals.
Do you have any experience with this new extraction option?.
Do any data on advantages / disadvantages?
Thank you again!
You can use TRANLOGOPTIONS to specify credentials for ASM user and also to specify whether GG has to use redo log or archive logs for fetching the changes. You can check this link for further details.
Then the capture process transactions is made from a single node of the Cluster and understand that the system generates Trail files with the transactions ordered by SCN, then the application at the target are consistent.
Reading some documentation on installation of GG in RAC, and to support the process of Extract failover if the primary node fails capture, we must use some file sharing mechanism between the nodes of the cluster shared files trail, checkpoint configuration.
It seems that ACFS is not supported and must be used DBFS?
Is it?Thanks.Best regardsArturo
Oracle GoldenGate can, at any time, runs on a single server. So which means, if one node goes down OGG should start from other node. So for recovery, at minimum, checkpoint files and trail files have to be stored at cluster independent of server availability. You could read this Oracle whitepaper to get a clear idea.
I would also recommend that you use the Integrated Capture (if you are on Oracle 188.8.131.52 or higher) as this makes the management of nodes going up and down automatic. If you use the classic Extract, and a node is added or removed, you'll need to manually modify you parameter file and possibly rebuild extract.
Also we have the option to install HA Application Agents on top of CLusterware to manage the failover of Golden Gate instances starting with Oracle 184.108.40.206.
Take a look at:
Thanks for the help.