Information technology — Power efficiency measurement specification for data center storage
|Publication Date:||1 November 2019|
|ICS Code (Information technology (IT) in general):||35.020|
This document describes a standardized method to assess the energy efficiency of commercial storage products in both active and idle states of operation. A taxonomy is defined that classifies storage products in terms of operational profiles and supported features. Test definition and execution rules for measuring the power efficiency of each taxonomy category are described; these include test sequence, test configuration, instrumentation, benchmark driver, IO profiles, measurement interval, and metric stability assessment. Qualitative heuristic tests are defined to verify the existence of several capacity optimization methods. Resulting power efficiency metrics are defined as ratios of idle capacity or active operations during a selected stable measurement interval to the average measured power.
This document defines methodologies and metrics for the evaluation of the related performance and energy consumption of storage products in specific active and idle states.
Storage products and components are said to be in an "active" state when they are processing externally initiated, application-level requests for data transfer between host(s) and the storage product(s). For purposes of this document, idle is defined as "ready idle", in which storage systems and components are configured, powered up, connected to one or more hosts and capable of satisfying externally initiated, application-level initiated IO requests within normal response time constraints, but no such IO requests are being submitted.
This document addresses storage products supporting block or file data access. Block access and file access refer to the type of service provided typically by Storage Area Network (SAN) and Network Attached Storage (NAS) systems, respectively. It is not appropriate to use this document to ascertain power efficiency for anything other than these two access modes. This document includes:
• A generalized taxonomy for storage products (clause 5);
• An assessment mechanism for software-based Capacity Optimization Methods (clause 6);
• Measurement and data collection guidelines for assessing the power efficiency of block- and filebased storage products in both active and ready idle states (clause 7);
• Metrics describing storage product power efficiency (clause 8);
• Required disclosures for a test result published as a SNIA Emerald™1 Power Efficiency Measurement test result (clause 9 ).
The purpose of a SNIA Emerald™ Power Efficiency Measurement is to provide a reproducible and standardized assessment of the energy efficiency of commercial storage products in both active and ready idle states.
1. Tested systems shall be comprised of commercially released products and components;
2. Tested systems shall employ settings, parameters, and configurations that would allow end-users to achieve power efficiency levels equivalent to the published result;
3. All data published as an SNIA Emerald™ Power Efficiency Measurement test result shall be gathered from test execution conducted according to this document;
4. Test execution shall complete in its entirety and without test failure or test error messages;
5. Software features which invoke, generate, or use software designed specifically for the test shall not be used. Configuration options chosen for test execution shall be options that are generally recommended for the customer;
6. the equipment shall be provided by the power mains that are being monitored (not by an internal UPS).
A SNIA Emerald™ Power Efficiency Measurement shall be a good faith effort to accurately characterize the power requirements of the tested system. The precise configuration used in a SNIA Emerald™ Power Efficiency Measurement is left to the sponsor of a test. Any commercially released components may be used, and a focus on new or emerging components or technologies is encouraged.
A SNIA Emerald™ Power Efficiency Measurement test result provides a high-level assessment of the energy efficiency of the tested system in specific ready idle and active states. It is not an attempt to precisely model or reproduce any specific installation.
Actual performance and energy consumption behavior is highly dependent upon precise workload, environmental and usage parameters. While a SNIA Emerald™ Power Efficiency Measurement test result is intended to provide a realistic and reproducible assessment of the relative power efficiency of a system across a broad range of configurations and usage patterns, it cannot completely match the precise needs of any one specific installation.
1 SNIA Emerald™ is a trademark of the Storage Networking Industry Association. This information is given for the convenience of users of this document and does not constitute an endorsement by ISO of the product named.