Installation, Deployment, and Initial Configuration
This domain covers taking a racked PowerStore system to a working cluster: cabling, discovery, the Initial Configuration Wizard, network planning, licensing, and the services that must run before the array is useful to hosts.
Deployment sequence
- Rack the base enclosure and any expansion enclosures, connect both power supplies to independent PDUs, and cable each node's management port to separate management switches.
- Cable data ports to the intended SAN and LAN switches; do not leave one node unconnected.
- Power on and discover the unconfigured system with the PowerStore Discovery Utility or by connecting to the node service port. The Initial Configuration Wizard sets the cluster name, management addresses, DNS, NTP, and administrator credentials.
- Configure storage networks for iSCSI and file, register the license, enable SupportAssist, CloudIQ, and alert notifications, then verify health before adding hosts.
Network and service rules
- Management, storage, and file/NAS traffic should be separate subnets or VLANs. Mixing iSCSI traffic onto the management network is a classic design error.
- NTP must be correct before AD/LDAP integration, SMB domain joins, or Kerberos NFS; clock skew breaks authentication even when the network is fine.
- DNS is required for SMB and for resolving NAS server names; plan forward and reverse records.
- Each node needs its own management address, and the cluster needs a floating management address.
Worked example
A two-appliance deployment must serve iSCSI to VMware and SMB to Windows clients. Plan four node management addresses plus the cluster address on the management VLAN; two storage VLANs with one interface per node for iSCSI; and NAS server interfaces on a file VLAN. Configure NTP and DNS first, then create the NAS server and join AD, because the SMB share cannot be created until the NAS server has an interface and a working domain relationship.
Traps that cost marks
- Cabling only one node to the management or data network, creating an avoidable single point of failure.
- Attempting an AD join before NTP and DNS are correct.
- Using the management network as the iSCSI or NAS data network.
- Skipping the post-install health check or the SupportAssist connection, then discovering the gap during an upgrade.