NREL Plans Strategies for Green Data Centers July

NREL Plans & Strategies for Green Data Centers July 15, 2008 Kevin Donovan, Data Center Manager National Renewable Energy Laboratory NREL Energy Efficiency and Renewable Energy Technology Development Programs Efficient Energy Use • Vehicle Technologies • Building Technologies • Industrial Technologies Renewable Resources • Wind • Solar • Biomass • Geothermal • Renewable Hydrogen Energy Delivery and Storage • Electricity Transmission and Distribution • Alternative Fuels • Hydrogen Delivery and Storage 3 Foundational Science and Strategic Analysis Influences to Plans and Strategies for Green Data Center • Access to renewable energy – 750kW PV array – Renewable Fuels Heating Plant – Renewable Energy Credits • World-class researchers in NREL’s Center for Buildings and Thermal Systems • NREL experience building LEED Platinum facilities • Industry best practices for green data center management Green IT Data Center • Energy Efficiency • Operational Procedures • Management Practices Suppliers • Energy-Efficient Products • Sourcing Practices • Environmental Track Record Workplace • Desktop Footprint • Remote Access • Printers • Recycling & Disposal Policies • Corporate Culture Current Data Center • • • • 30 years old 2,500 sq. ft. Supports corporate data center functions and HPC Challenged to support growing demands for space, power and cooling • Inefficient power and cooling systems • No backup generator power • Recently completed an energy audit to identify prudent investments to extend longevity Research Support Facility (RSF) • LEED Platinum certified • 210,100 sq. ft. office space • Supports: – 800 administrative staff – 2,500 sq. ft. data center • Expected completion: June 2010 • Corporate data center functions to support > 2,000 staff • Entire building will be powered by renewable energy Energy Systems Integration Facility (ESIF) • Minimum LEED Gold certified • 130,000 sq. ft. • Supports: – 150 research staff – 15,000 sq. ft. High Performance Computing data center • Expected completion: December 2011 Why is Data Center Energy Efficiency Important? • Many data centers are facing a power and cooling crisis • 2-3% of all power produced in the U.S. is used in data centers • Data center power demands are expected to increase • Energy costs for equipment are expected to exceed initial equipment costs over lifecycle • Energy costs are rising Expedition to Mount Everest • Requires preparation, planning and conservation of resources • Strategize • Setting up fixed ropes • Movement of Teams • Logistics • Staging materials & equipment • Contingencies • • • • Delays due to weather Location of medical team Location of emergency oxygen Evacuation of injured climbers Expedition Management on Mount Everest • Best practices – Preventive maintenance • Inspect equipment before departure • Take spare parts and repair kit • Check battery fluid levels – Capacity Utilization • Plan your fuel usage • Send emails once daily for entire team via SAT • Limit Non-Emergency Satellite Phone Usage – Measurement • Monitor daily fluid uptake • Monitor utilization of fuel, battery power and water – Leverage available resources • Purchase local items – Innovation • Find creative ways to solve problems and conserve resources The Green Data Center Data Center Power Usage The data center energy challenge affects both physical data center and IT infrastructure % of total data center electricity use 35 30 25 20 15 10 Cooling systems Electrical and building systems 5 Chiller/ Humidifier AirCooling Conditioning PDU UPS Switch/ gen Lighting Information Technology Power use Optimize Data Center Infrastructure Energy Efficiency Source: Gartner Active Energy Management Optimize IT Infrastructure Energy Efficiency Equipment Strategies • Move from “always on” to “always available” model • Use energy-efficient equipment • Consolidate & virtualize • Right-size • Design for scalability • Proactive capacity management V Hig irtual h E i za t ffic ion ien cy Se rv e rs Storage Virtualization High Efficiency Energy Efficient Equipment es urc o es on f R ucti o ed ng oli ble R o a P C N o rk etw Servers • Server consolidation/ virtualization ratio: 6-8 to 1 • VMware and Solaris containers used for virtualization • Energy efficiency considerations: – Nominal increase in energy consumption for a higher server utilization – Highly efficient server power supplies are critical • Moving towards high-density blade servers • Server utilization goal: > 60% • Our challenge: managing server sprawl Storage • • Right-sized through use of thin provisioned storage Storage architecture matches performance and redundancy with requirements – – – Bulk data Mid-Range High-performance Ò Ò Switched SAN • • • Designed to scale quickly and easily “Pooled” storage increases better resource utilization Backup only what is required – – – Reduces backup hardware Reduces number of tapes Reduces energy consumption and waste disposal Virtual Pool of Storage Free Space • Storage utilization goal: 75% Network • Network must be “always on” • Right-size – 100Mbps to desktop – Reevaluate VoIP phone selection – Carefully manage # of ports • Consider ways to leverage network to improve environmental impact • Use virtual connections for blade servers Data Center Power Systems Strategies • Use renewable energy sources • Right-size • Design for scalability • Use energy-efficient equipment • Proactively manage Uninterruptible Power Supply (UPS) • Designed to provide redundancy • Few minutes of uptime • > 92% energy efficient • Operate in efficiency “Sweet Spot” Backup Power • Bio-diesel generator to provide backup power for building • Hydrogen fuel cell to provide backup power for data center – – – – – – 30KVA Works in conjunction with UPS Low maintenance Reliable: no moving parts Quiet Low environmental impact Proactive Power Management • Must track energy use of data center power systems • Data Center Efficiency metrics will be used to measure and track the energy efficiency of power and cooling systems • Data center energy costs must be considered in calculations Cooling System Strategies • • Use Hybrid cooling system Contain and recycle heat resource for building use Use super energy-efficient equipment Right-size & design for scalability Run data center five degrees warmer Run chilled water warmer Manage proactively • • • • • Hybrid Cooling System • Free cooling making use of a waterside, air or propylene glycol economizer • In-row cooling for high density racks • Contained pressurized floor cooling for low density racks • Raised floor design for maximum flexibility Recycling Heat • Layout data center to use well-defined hot and cold aisles • Contain heat produced by equipment • Explore use of data center waste heat for other building purposes Proactive Cooling Management • Must track energy use of data center cooling systems • Instrumentation • Regular reporting • Cooling trend analysis • Smart cooling Data Center Efficiency (DCE) Cooling + Power + Equipment DCE = Equipment Typical Target Expected Reduction in Data Center Power • 1.4 DCE • 30% reduction in power for equipment • 70.4% overall reduction in data center power The Green Workplace Source: Corné de Graaf RSF Desktop Environment • Manage workstation lifecycle for environmental impact Workstation Type Power Reduction • Use energy-efficient LCD monitors Standard PC • Use laptops with docking stations for road warriors & day extenders Laptop • Evaluating the use of thin-client technology at the desktop – Use VMware’s VDI – Desktops VM will run on data center servers – ~60 desktop VM’s /server Energy Star PC 400w 300w 0% 25% 60w <30w 85% 92% Thin Client Dynamic Workload Management • Server resources go from “always on” to “always available” • As resources are needed, they are brought online • As resources are no longer required, they are powered down or put in standby • Saves energy Printer Management • Multifunction energy-star 4.0 compliant devices • Reduce number of network printers by 75% • Strong justification will be required for personal printers • Do not use banner pages • Default to duplex printing • Business processes are becoming increasing paperless • Use recycled materials Role of Our Suppliers Partnerships with Our Suppliers • Design and implementation of state-of-the-art cooling and power systems • Application migration to virtual environments • Tools and techniques to manage dynamic workloads • Tools to track and monitor data center power, cooling, server and storage resources • Development of products that can be powered off or put in standby mode • Challenge to our suppliers: – More software to run in a virtualized environment – Hardware to be made more energy efficient – Hardware to be able to endure higher heat loads Best Practices for Green Data Centers • • Use energy-efficient equipment Upgrade legacy equipment • • • Consolidate & virtualize servers Right-size IT infrastructure Good relationship between IT and Facilities Best Practices for Green Data Centers (cont.) • • • Track & manage data center energy consumption Perform long-term capacity planning for power Make energy consumption a part of TCO analysis • • Implement targeted/adaptive cooling solutions Improve airflow management The U.S. Department of Energy’s National Renewable Energy Laboratory http://www.nrel.gov Golden, Colorado 36