Methods to identify, handle and recover from suspect SSDS in a clustered flash array

摘要: A technique predicts failure of one or more storage devices a array serviced by system and for establishes threshold conditions replacing the devices. The predictive periodically monitors soft hard failures (e.g., from Self-Monitoring, Analysis Reporting Technology), as well various usage counters pertaining to input/output (I/O) workloads response times heuristic procedure may be performed that combines monitored results calculate predicted recommend replacement devices, using thresholds based on current patterns In addition, policies provided in cost-effective manner ensures non-disruptive operation and/or SSDs, while obviating potential catastrophic scenario