Best Practices in NVMe Power Management

Recently, SearchStorage.com consulted Phison on the importance of NVMe SSD power management. NVM, or non-volatile memory, is highly relevant to many businesses, including manufacturers of industrial, automotive, consumer/PC, and enterprise-level storage products. This post describes how Phison approaches power management to satisfy the unique requirements of devices and workloads.

The difference between client and enterprise SSDs

NVMe power management is related to power consumption, device efficiency, performance, reliability, and thermal management. SSDs are designed to adjust power usage to meet the desired access patterns and maintain system temperature.

Client SSDs

Client (consumer) and enterprise SSDs are very different, and that difference applies to power management as well. Client SSDs tend to be idle or non-operational most of the time. Power management features for client SSDs are designed to transition the drive to low power states to help achieve longer battery life.

The Autonomous Power State Transition (APST) NVMe power management feature is typical for client SSDs. APST lets the host set policies for when the device transitions to, and wakes up from, a low-power state. The device should transition to lower power mode when non-operational for a specific time set by the host. If the device does not transition to low-power mode, it continues to draw active idle power, and this will quickly drain the battery in laptops and notebooks.

The choices and expectations about what power state policy to use, the amount of idle time before transitioning to lower power, the power drawn when idle, and other settings also differ for desktop PCs and laptops, as well as “ruggedized” notebooks used in harsh conditions. Their airflow and cooling vary, too, with notebooks offering far less airflow than desktops. Phison works with these vendors to deliver the most appropriate power management for their specific product design.

Enterprise SSDs

Enterprise and data center NVMe SSDs have far more demanding jobs to cope with than client SSDs. Enterprise and data center SSDs are active most of the time, if not all the time, and preventing them from overheating relies on effective power management. When the system’s airflow is not enough to cool the SSD, power management features are used to “throttle” performance to lower device temperature. Otherwise, the temperature will continue to rise, leading to thermal shutdown. With effective power management, a device can operate in full performance mode by keeping a device’s temperature within the normal range.

Striking a balance between power, performance, and thermal conditions is critical for a drive’s reliability. Improper power management in the worst case can lead to data loss.

Read: AHCI vs. NVMe

The tradeoff between power savings and latency

While power efficiency and conservation are important, there is a catch. NVMe power management is critical for different levels of power-saving and thermal conditions, but it also comes with a tradeoff: latency. The time it takes the drive to exit a low-power mode — to “wake up” — can be appreciable, and so can the time it takes to enter low-power mode. Plus, the lower the power state, the longer it takes to sleep and wake up the drive. Device manufacturers need to consider impact of enter and exit latency on performance and response time, which could affect user experience.

NVMe power management methods are directly related to each SSD’s intended function and thermal requirements, which vary between client and enterprise SSDs and from one industry to another. Military and aerospace technology demands can be vastly different from healthcare devices or PC gaming, and power management directly corresponds to performance and user experience. Phison partners can feel sure that NVMe SSDs for their products are application-aware, optimized for their intended workload, with the right power control features for their needs.