In iCluster, before considering adding another journal…
As a follow-up for one last time… to more journals vs. fewer journals, if you do decide that you want another journal to help with latency, stop before you add it and consider the following.
- Confirm that objects currently being journaled are part of replication and are needed.
- Eliminate temporary objects from journaling.
- Try the iCluster function to Mirror Contents *NO for files that must be over on the backup node but contents are temporary.
- Work with your application experts to help you reduce the number of work files being created by the process as well as how these are created and cleared at the end of the process.
- Try turning on some supported journal options (at the minimal SYSDFT – RMVINTENT, MAXOPT2).
- Try the Optimize Apply for Updates *YES for any groups with apply latency.
You may be able to resolve your latency issue without the need to add another journal!
[box type=”info”]Learn more about iCluster high availability[/box]