PowerStore Architecture and Hardware Components
This domain tests whether you can describe a PowerStore system from the enclosure up: what an appliance contains, how two nodes share NVMe drives, how a cluster scales, and which model families can coexist. Exam items often give a topology or model number and ask what can be added, failed over, or mixed.
The appliance is the unit of scale and failure
- A PowerStore appliance is a pair of active-active nodes in a 2U base enclosure that holds up to 25 2.5-inch NVMe SSDs.
- Both nodes serve I/O simultaneously; neither is a passive standby. If one node fails, its peer takes over the affected volumes and paths.
- Drives are dual-ported NVMe attached directly to both nodes. There is no SAS expander or traditional RAID controller in the data path.
- Each appliance supports up to three 2U, 24-drive expansion enclosures, so capacity grows without adding nodes.
- A cluster contains up to four appliances managed through one PowerStore Manager, and volumes can be migrated between appliances non-disruptively.
Model families and naming
PowerStore T models are unified: block (FC, iSCSI, NVMe-oF) plus file (NFS, SMB). PowerStore X models run appsON with an embedded vSphere layer. The digits indicate a performance tier, not a protocol or drive type; every model is all-NVMe. T and X appliances cannot be mixed in one cluster.
Worked example
A customer has a PowerStore 1000T with two expansion enclosures and asks for more capacity with minimal disruption. The first option is a third expansion enclosure, which is added online and keeps the same two nodes. The second is adding a second appliance to the cluster, which adds two nodes and a separate failure domain; existing volumes can then be migrated to it. Adding an expansion enclosure does not add a node, does not change the cluster appliance count, and does not create a new failure domain.
Traps that cost marks
- Saying an expansion enclosure adds a node or a cluster member.
- Believing one node is idle until failover; both are active.
- Mixing T and X in one cluster, or claiming a cluster can hold more than four appliances.
- Treating the model number as a protocol indicator, or assuming any model uses SAS drives.