Storage Management, Monitoring and Troubleshooting
This domain tests daily operations: where to configure, what to monitor, and how to isolate a fault methodically.
The management surface
Huawei arrays are managed through DeviceManager (web GUI), a CLI over SSH, and one-click information collection for logs; alarms and events also stream via SNMP traps or syslog, and health reporting can be forwarded through the remote maintenance service. Configure NTP first, then AD/LDAP authentication, RBAC roles and audit logging, so later troubleshooting has a trustworthy timeline and an attributable change history.
What to watch
Front end: port bandwidth and IOPS, host queue depth. Controllers: CPU and cache utilization, cache hit ratio, write-back versus write-through state, back-end port load. Disks: utilization, latency, and any reconstruction in progress. Capacity: pool usage, snapshot space, thin-provisioning overcommit. Replication: link state, delay and RPO lag. Degradation is usually one saturated resource, so read metrics along the I/O path and stop at the first one that is pegged.
Worked example
An OLTP database jumps from 1 ms to 12 ms. Check in path order: front-end port throughput is at 95% while controller CPU is 60% and disks sit at 40% - the fabric edge is saturated, not the backend. Confirm on the host that all paths are active/optimized and that queue depth is not inflating latency, then rebalance by adding front-end ports or redistributing hosts through the multipath policy. Adding disks would not help.
Traps
Rebooting or reconfiguring before collecting logs; clearing alarms without finding the root cause; blaming the array for latency caused by host queue depth or a switch ISL; ignoring snapshot-space alarms until writes fail; correlating events across devices without synchronized clocks; overlooking audit logs when explaining who changed a mapping.