Adaptive data protection for a multi-node data cluster

Abstract

An illustrative data storage management system protects data in a replicated multi-node system ('cluster'). The illustrative system creates a new administrative entity (“pseudo-client”) that represents the cluster within the system. The system intelligently performs as an adaptive “set it and forget it” system. Preferences for the pseudo-client, including alternate rules that cover various contingencies in the cluster, need not be reset as the cluster undergoes node failures and/or role changes over time. Rather, the disclosed system adapts, on each successive secondary copy job, to the current configuration of the cluster without requiring a system administrator to provide further input or analysis. This approach ensures that jobs can begin promptly as scheduled or on demand. Component selection is based at least in part on geography/zone and/or network topology considerations, which improves the likelihood of job completion and success, and RPO/RTO service levels are more likely to be met.