The Oracle Database patchset software (for example, 11.2.0.3) requires the Exadata-specific version.

Rolling patching is faster than non-rolling patching, but database down time is longer.

The Oracle Database software on Exadata is updated using standard OPatch and the Oracle Universal Installer.

Running Exadata with different storage server software versions is supported, but should be minimized to rolling patching scenarios.

The database server operating system software (other than kernel and OFED) may be installed or updated from non-Exadata sources

Storage server updates require access to an Unbreakable Linux Network (ULN) based repository.

High redundancy provides more usable storage.

Normal redundancy provides more usable storage.

High redundancy gives you better I/O performance.

High redundancy provides High Availability (HA) during disk failure and rolling Storage Server patching

Normal redundancy provides HA during rolling Storage Server patching.

Instance caging is the right approach to make sure that database instances are not stealing memory from each other.

The CPU Resource Manager and IORM need to be enabled to make instance caging to work as expected.

Database instances need to be restarted to make changes to cpu_count effective.

For a system with 16 threads, the sum of cpu_count for all databases should notbe higher than 32 to prevent oversubscribing

The init.org parameter resource_manager_plan should be set.