Installation, Cabling, and Initial Configuration
This domain covers the physical build and first-boot tasks: racking, cabling both nodes, addressing management and storage networks, and creating a cluster that later appliances join.
Address planning before power-on
- Plan three address types per appliance: a management IP for each node, an appliance/cluster management IP, and storage IPs for iSCSI or NVMe/TCP, one per node port used for host I/O.
- File configurations need additional NAS server IPs on the file network.
- NTP and DNS must be reachable at build time. Time skew breaks replication, certificate validation, and log correlation; DNS is required for SMTP, SupportAssist, and NAS name resolution.
- Discovery is done with the PowerStore Discovery Tool, or by browsing to the node address from a laptop placed on the same subnet.
Cabling rules
- Connect both nodes to the management switch. A single management link is a classic build defect that hides one node from monitoring.
- FC front-end ports go to SAN switches and are zoned afterwards; Ethernet front-end ports go to top-of-rack switches.
- Give iSCSI its own VLANs rather than sharing the management VLAN, and keep MTU consistent end to end.
- Inter-appliance (cluster) traffic must stay on a low-latency, non-routed path; do not stretch it across a WAN.
Initial Configuration Wizard
- Accept the EULA and set the admin and service account passwords.
- Enter node management IPs and the cluster management IP.
- Configure DNS, NTP, and optionally SMTP, SNMP, and syslog.
- Create the cluster on the first appliance, then use Add Appliance for each additional unit.
- Configure storage networks, then create NAS servers if file services are required.
Worked example
Management VLAN 10.0.10.0/24: node A takes .20, node B takes .21, and the cluster takes .22. Storage VLAN 10.0.20.0/24 carries iSCSI portals .30 and .31. Reasoning: the cluster IP is what administrators and replication partners reach, so it must sit on the same L2 network as the management workstations; the storage portals must be isolated from it so host I/O does not compete with UI and call-home traffic.
Traps
- Leaving the array reachable only on the service port instead of the customer management network.
- Skipping the second node's management address, which blinds you to node-level alerts.
- Mismatched MTU or VLAN tagging between host NICs and array ports.
- Deferring SupportAssist, which later blocks call-home and remote troubleshooting during an incident.