Drive Energy Management for Extending SSD Lifespan Through Smart Power States

Abstract

Introduction: The increasing demand for high-speed and energy-conscious SSDs has elevated the importance of intelligent power management in storage architectures. This paper investigates how modern NVMe SSDs manage dynamic power while operating ov        er PCIe. As workloads become more varied and intensive, controller behavior plays a key role in balancing performance and power. Managing energy efficiency without compromising throughput has become a key challenge in system-level design

Objectives: This work aims to analyze how NVMe-defined power states influence performance and energy efficiency. NVMe SSDs support multiple power states (NPSS = 4), ranging from 20.00W in PS0 to 12.00W in PS4, while maintaining a fixed idle power of 5.00W. The objective is to determine how power state transitions affect overall system behavior under diverse load conditions.

Methods: Using standard NVMe admin commands (feature ID 0x02) along with OCP-specific extensions (feature ID 0xC7), power state transitions were initiated from the host side. These commands enabled state changes (e.g., from 0x14 to 0x10) without halting ongoing drive operations. A workload-driven testing methodology was applied, where SSDs were benchmarked under controlled performance scenarios during and after transitions.

Results: Ege Results show a consistent 100µs latency overhead for all power state transitions. Measured throughput and latency metrics revealed linear scaling across the defined power states. These findings validate a predictable trade-off between power and performance, providing quantitative insights into real-time SSD behavior. Thermal stability and controller activity remained within operational limits throughout all transitions.

Conclusions: This study presents key findings for SSD firmware optimization targeting power-aware systems. Practical strategies are proposed to balance performance and energy savings in enterprise and consumer environments. By enabling smarter transitions between power states, firmware can maintain responsiveness while reducing energy draw. These insights support better controller design and future power management strategies in NVMe-based architectures.