Why and how to implement NVMe power management

Fast, reliable NVMe SSDs have revolutionized data storage. There’s a downside to NVMe SSD technology, however: high power consumption. Fortunately, the NVMe specification offers numerous power management features. Here’s what you need to know.

NVMe power management is necessary to achieve an acceptable balance between platform thermals and total power consumed by the SSD. Even though the form factor defines maximum power for an SSD, the host can use NVMe power states to change the amount the SSD consumes, said Imran Hirani, director of product architecture for Taiwan-based enterprise SSD manufacturer Phison Electronics Corp. That change is based on the desired access pattern, as well as the maximum available thermal solution and airflow for the system.

KEY POWER MANAGEMENT TECHNOLOGIES AND FEATURES

A host can access NVMe power management features in three different ways:
• To set the power state conditions for the Autonomous Power State Transition feature,
use the set feature command (FID=0xC, APST). The SSD will then transition to different
power states based on the set condition.
• Use the set feature command for power management to change a current SSD power
state.
• Use the set feature command for host-controlled thermal management to establish
two temperature thresholds. The device will automatically transition to the lower
power state once it reaches the set temperature.

NVME POWER RISKS

NVMe power management features find acceptable balances between power, performance,
product reliability and customer experience. However, there are risks if organizations don’t
properly configure NVMe power management.

They include the following:
• Failure to properly manage the SSD’s active power can lead to excessive heat
generation, which potentially leads to a level where available system airflow will not
be able to sufficiently cool the SSD. “In worst cases, the SSD can reach its thermal
shutdown limit and will shut down,” Hirani said.
• When in the nonoperational power mode, a drive that fails to transition to lower
power modes will continue to draw power even through the drive is idle. “This will
drain the battery in the laptop/notebook faster, leading to a poor customer
experience,” Hirani warned.

• While NVMe power states provide various power-saving levels, there’s also a tradeoff
in the time it takes to enter and exit each state. “The lowest power state will take the
longest to enter and exit,” Hirani said. “Not considering the enter and exit latency
could lead to lower performance or longer response time.”

SCALING THE NVME STORAGE CONTROLLER

The current most common shared storage system is the active-active controller. Two
controllers each have access to the other’s drives in case one controller fails. Two controllers
do not generally increase the number of fully addressable NVMe drives. This is to prevent
huge performance degradation when one NVMe storage controller is unavailable. Scaling
the storage controller number beyond two requires a bit more creative architecture.

Scale out the number of controllers. Block storage scale-out is typically through some type of
clustering. Scale-out for file and object storage is generally a global namespace, but not
always. All scale-out is either shared everything or shared nothing architectures. Block is
more likely to be shared everything, but not always. Shared nothing is more common for file
and object.