Ecodesign Technical Assistance Study on Standards for Lot 9 Enterprise Servers and Enterprise Data Storage - Final Report (Intertek, June 2016)

Overview

The study provides technical assistance to standardisation tasks supporting a potential future EU Ecodesign Regulation for Lot 9 equipment (enterprise servers and data storage). Its aim is to lay the foundation for robust, durable standardised methods of measuring the energy efficiency of servers (especially rack and blade servers) and storage devices; measurement approaches are needed so manufacturers can assess compliance and national bodies can assess market compliance. The report was written by Intertek, is dated June 2016 and is current as of April 2016.

On policy context, the 2012-2014 Working Plan identified enterprise servers and data storage as a key product area with initial estimated savings of 135 PJ/year as of 2030. The preparatory study was completed in September 2015. The Commission had already regulated only power supply efficiency in a subset of servers through Regulation (EU) 617/2013 of June 2013. The parameters considered include active and idle power demand, energy-proportional operation, power supply efficiency and power factor, operating temperature and humidity, acoustic noise, removability of components, data sanitisation and critical raw material content. The saving potential for servers and storage was estimated at about 17 TWh by 2030.

The gap analysis reviews the Commission requests M/545 (computers), M/462 (telecom infrastructure) and M/543 (material efficiency) and standards such as the CENELEC EN 50600 series, ISO/IEC 30134-4 and -5, SPEC SERT v1.1.1, SPECpower_ssj2008, SNIA Emerald, ASHRAE TC 9.9 thermal guidelines, NSF/ANSI 426 and IEEE 1680.4, and gives a traffic-light summary of gaps for servers and storage. Conclusion: there is much standardisation activity on server standards, while storage energy efficiency measurement standards are less advanced.

SERT is SPEC's server efficiency rating tool: it is designed for reproducibility, fairness, verifiability and usability, is hardware and OS agnostic, tests servers 'as shipped' and targets a run time of about five hours. 'Worklets' test components such as processors, memory and storage separately, with results normalised against an arbitrarily selected baseline server. Licence cost ranges from $900 (not-for-profit reduced rate) to $3,000. The report's limit is its date: it is from 2016 and does not cover the later regulation (2019/424); in addition no stakeholder took part in the SERT beta programme.

Key points

1. Saving potential of ecodesign measures for servers and data storage was estimated at about 17 TWh by 2030; the Working Plan initially estimated 135 PJ/year for 2030.
2. Parameters considered include operating conditions (temperature and humidity) and material efficiency (reuse, recycling), as well as active and idle power, energy proportionality, power supply efficiency and power factor, acoustic noise, data sanitisation and critical raw material content.
3. Standardisation requests reviewed: M/545 (computers), M/462 (telecom infrastructure) and M/543 (material efficiency).
4. Standards reviewed include the CENELEC EN 50600 series, SPEC SERT v1.1.1, SNIA Emerald and ASHRAE TC 9.9 2011 thermal classes; a traffic-light figure summarises standards gaps for enterprise servers and data storage.
5. SERT tests servers 'as shipped' with a target run time of about five hours; results come as XML and HTML/TXT; the SPEC licence costs $900 to $3,000 (normally $2,800).
6. A wider ASHRAE temperature range can lower data centre cooling cost, but since the same heat must still be removed, energy use can shift into the server, requiring larger, more energy-consuming fans and larger heatsinks.
7. Server idle/active metric work continued in 2016 with SPEC and ENERGY STAR; server KPIs and ITEE in ISO/IEC 30134-4 and EN 50600 were expected in early 2017; storage metrics were expected after 2016.
8. The only existing regulation before an ecodesign measure was Regulation (EU) 617/2013 of June 2013, which covers only power supply efficiency in a subset of servers.
9. Despite repeated publicity no stakeholder took part in the SERT beta programme, possibly because most are already involved in SERT development; the trial licence was nevertheless useful for the project team's own testing.
10. Metrics analysis used thirteen servers, comparing the effect of CPU, RAM and number of disks on power.

Using it in design

• When buying servers and storage use SERT results and ENERGY STAR server specifications to compare energy efficiency; the EU Code of Conduct also refers to ENERGY STAR for procurement.
• When choosing an ASHRAE thermal class for IT equipment include in the total energy calculation that a wider temperature range can cut data centre cooling cost while raising in-server fan and heatsink energy.
• Besides PUE plan measurement infrastructure for IT workload efficiency metrics (ITEE and ITEU, ISO/IEC 30134-4 and -5); these standards were being prepared for server KPIs.
• In server specifications ask explicitly for power supply efficiency and power factor, idle power and energy proportionality; Lot 9 treats these as key parameters.
• For storage equipment, knowing that no mature standard measurement method exists yet (SNIA Emerald, ENERGY STAR Storage after 2016), ask suppliers to document measurement method and conditions.

Who should read it

Data centre designers and procurement teams selecting server and storage hardware, measuring IT load efficiency and following the related EU standardisation work.

Document contents

1. 1. Introduction
2. 2. Policy context: 2.1 European initiatives, 2.2 International and industry initiatives
3. 3. Standardisation context: basic principles, entities involved, the standardisation process
4. 4. Product scope: enterprise servers, enterprise storage, definitions
5. 5. Performance parameters considered
6. 6. Relevant EC standardisation requests: M/462, M/543
7. 7. Standardised test method gap analysis on enterprise servers and enterprise data storage
8. 7.1 Relevant European standards: CENELEC EN 50600 series
9. 7.2 Relevant international standards: ISO/IEC 30134-4/-5, IEC TR 62635, ISO 7779
10. 7.3 Relevant national standards: PAS 141, NSF/ANSI 426, NIST SP 800-88
11. 7.4 Industry standards: SPECpower_ssj2008, SERT v1.1.1, SNIA Emerald, ASHRAE TC 9.9, ECMA-74, IEEE 1680.4
12. 8. Standards engagement and SERT v1.1.1 beta testing
13. 9. Conclusions
14. Appendix 1 Coverage and status of key standards / initiatives
15. Appendix 2 Server and Storage Standards Listing
16. Appendix 3 Practical Insights on SERT testing
17. Appendix 4 Metrics Paper

Related standards

• Regulation (EU) 2019/424
• EN 50600
• SERT
• SNIA Emerald
• ASHRAE TC9.9

Read the document

Publication details

Authors

Catriona McAlister; Anson Wu; Jonathan Wood; Bob Harrison; Stephen Fernandes; Fiona Brocklehurst

Publisher

European Commission, DG GROW (Intertek)

Year

2016

Document number

ET-02-16-987-EN-N

Document language

English

Source type

Government / public

Access

Open licence — downloadable

DOI

10.2873/782959

Licence

EU reuse (2011/833/EU) — reuse permitted with acknowledgement of the source

Licence evidence

op.europa.eu

Citation