DPM Host Power-Down Eligibility for Low-Traffic Workloads: An Unresolved Mode Decision

vSphere Distributed Power Management (DPM) can power down idle ESXi hosts in a DRS-enabled cluster, which may reduce power and footprint for low-traffic workloads. The unresolved decision is whether to run DPM in fully automated mode or manual recommendation mode when workload variability and SLA sensitivity are both present.

DPM depends on cluster-level DRS, shared storage, and VM mobility, and it only considers hosts that can safely enter maintenance mode. Hosts with local storage, hardware passthrough, or pinned VMs are generally unsuitable. The aggressiveness setting shifts the balance between power savings and consolidation risk, while automated power actions can conflict with capacity planning or licensing models that count powered hosts. DPM naming, defaults, and integration with DRS and vSphere Lifecycle Manager are version-sensitive.

Verification would involve checking the cluster's DRS and DPM configuration, reviewing DRS recommendation history, and confirming candidate hosts have no blocking dependencies.

Which signals distinguish a host that DPM can safely power down from one that merely appears idle? How should the aggressiveness threshold be selected when workload variability is high but SLA tolerance is low? Does the automated-versus-manual choice change across vSphere releases where DPM defaults or integration have shifted?