Verifying cluster health is the absolute prerequisite before attempting any upgrade or maintenance activity.
An upgrade involves taking nodes offline (rebooting) sequentially. When a node goes offline, the cluster relies on the remaining nodes to serve I/O and maintain data redundancy.
If the cluster is already in a Degraded state (e.g., a drive is failed, or capacity is nearly full), taking a node offline could result in Data Unavailable (DU) or Data Loss (DL).
PowerFlex Manager and the CLI provide "Pre-upgrade Health Checks." These checks validate:
No active rebuilds or rebalances are running.
All SDS and MDM components are connected and synchronized.
There is sufficient spare capacity to handle a node failure or maintenance mode.

To utilize Fault Sets effectively for their intended purpose (rack-level fault tolerance), the cluster architecture imposes specific minimums.
PowerFlex relies on distributing data copies (Mesh Mirroring). The standard redundancy scheme creates 2 copies of data. To ensure that data remains available even if one entire Fault Set (e.g., an entire rack) fails, the system must be able to distribute the mirrors across different Fault Sets.
Consequently, you generally need a minimum of 3 Fault Sets (System requires 3+ Fault Sets to support 2-copy layouts effectively). Since a Fault Set must contain at least one node, the cluster requires a minimum of 3 nodes to support a Fault Set configuration.

True: The specific purpose of Protected Maintenance Mode (PMM) is to ensure data availability.
How it works: When you enable PMM on a node, the system checks the data chunks residing on that node. Before the node is allowed to effectively "leave" the cluster (stop serving I/O), PowerFlex copies (mirrors) that data to other nodes in the cluster. This ensures that even when the node is physically removed or rebooted, the cluster retains the required number of data copies (redundancy), ensuring no data is unavailable (DU) or lost (DL).

* Group disks with similar performance characteristics (Option C): The primary architectural purpose of a Storage Pool is to aggregate physical drives that share the same performance profile (e.g.,
"All SAS SSDs" or "All NVMe").
* Performance Uniformity: PowerFlex stripes data across all disks in a pool. If you were to mix slow HDDs and fast SSDs in the same pool, the performance of the entire pool would be dragged down by the slower drives (the "straggler" effect). Therefore, pools are used to segregate media to guarantee consistent IOPS and latency for the volumes created within them.

MDM manages cluster metadata, configuration, and state. It ensures consistency across the system. It does not store actual data or present volumes.