I'm migrating to 2 node RAC 220.127.116.11 and wondering if ocfs2 is still best option for storing arch logs ?
I've found it pretty stable in 10.2.0.3 , but looks like there are some alternatives .
Could You elaborate on that subjects.
My redo rate is about 1 TB per day with 200gb per hour peak .
We have 18.104.22.168 3-Node RAC. Archive logs, RMAN backups etc.... are stored in Fast Recovery Area which is stored in ASM Disk Group consisting of multiple RAID10 SAN LUNs with many back-end spindles...write performance is real good. We backup archive logs from FRA to an ACFS file system on another ASM Disk Group which is on another set of SAN LUNs. We use Data Guard, so archive logs are backed up from the standby site.
OCFS2 is definitely not the best option (never has been for RAC) => since Oracle 10 it is directly in the ASM diskgroup / FRA.
One reason is: With OCFS2 you have another clusterware running on the node (besides Oracle clusterware), which is not directly integrated.
So ASM (in FRA) and maybe a copy in ACFS for easier administration is the recommended option.
OCFS2 requires an interconnect. Preferably private because it typically has lower latency.
So,my recommendation is ACFS, because is fully supported and meets all the needs for backup and recover without additional configuration outside the clusterware. Centralized management (such as rolling upgrade, add/remove nodes and so on).
So acfs is the choice , one thing which bothers me is maturity of acfs as its quite fresh .
Did You find it stable in production with 1TB per day of archive and I saw few space related bugs with acfs .