Paint IT Green A Guide to Lowering Your ICT Carbon from Data Centre to Desktop
Welcome To The University of Hertfordshire Professor Di Martin, Chief Information Officer
10:00Keynote Speaker: Dr Ian Bitterlin, CTO, Prism Power, 'Shared services': increased energy efficiency via consolidation in larger facilities 10:30Break & Refreshments 11:00Steve Phipps, Data Centres Manager, UH, A Best Practice Guide to Greening the Micro Data Centre 11:40Katherine O'Brien, Environmental Coordinator, UH, Useful Strategies for Improving Environmental Performance 12:00Pete Sands, Lead Data Centre Design Consultant, Future-Tech SCI, The Cost of Data Centre Ownership 12:30Lunch 13:30Howard Noble, Principle Investigator, Oxford University, "When Should I Do My Bit (Workshop) 14:00Steve Phipps, Data Centres Manager, UH, Providing a Green Benchmark 14:30Mark Johnson, Consultant, AEA Group, What CRC means for HE/FE/Public Sector and what to do about it 15:00Break & Refreshments 15:30Panel Discussion Chair: Professor Andrew Starr, Director of the Centre for Sustainable Communities, UH, Panel: Barry Lewington, 16:00Richard Smeeton, CTO, UH, Wrap Up of the Day 16: :00Networking / Refreshments Paint IT Green Conference Programme Richard Smeeton, Chief Technology Officer, UH
Next Presentation Keynote – Dr Ian Bitterlin
'Shared services': Increased energy efficiency via consolidation in larger facilities? Dr Ian F Bitterlin PhD BSc(Hons) BA DipDesInn MCIBSE MIET MIEEE MBCS MBIFM Prism Power Ltd, UK
Three steps to Sustainability Reduce consumption – The social & economic value of the data processed? Improve efficiency – Not just PUE but also IT software & hardware Use energy from renewable sources – Building a legacy datacentre next to a hydro- electric scheme is NOT a sustainable design, it is a waste of valuable green energy on an inefficient data centre
Value of the traffic? Rise of the Hyper Giants: Five years ago, Internet traffic was proportionally distributed across tens of thousands of enterprise managed web sites and servers around the world. Today, most content has increasingly migrated to a small number of very large hosting, cloud and content providers. Out of the 40,000 routed end-sites in the Internet, 30 large companies – hyper giants like Limelight, Facebook, Google, Microsoft and YouTube – now generate and consume a disproportionate 30% of all Internet traffic. Arbor Networks, 13 th October 2009
Targets should be IT & Cooling?
IT hardware efficiency = 0%? Worldwide chip utilisation = ~5% IT power = chip, memory, drives, I/O, power conversion & fans At ~5% IT load the average server draws >60% power
Where the power goes … PUE=1.6 IT load Distribution & UPS losses Cooling fans, pumps & compressors Lighting & small power Security, BMS Ventilation – Fresh Air Communications 5 kW 15 kW 240 kW 500 kW 35 kW 3 kW 2 kW Total 800 kW 1MVA PUE = 800kW/500kW = 1.6
What is the PUE?
Key factors for high efficiency High efficiency servers & software Heavy Virtualisation – Multiple applications per server High load Vs capacity, M&E Free-cooling or Fresh-air cooling Relaxed set-points for temperature + humidity Strict air-flow management High efficiency UPS
Free-Cooling – not free, but cheap
Free Cooling Vs temperature
Electrical systems: Efficiency Vs Load
Easy to do on a large scale Leverage infrastructure, NOC etc Modularise Consolidate the load(s) Build rooms and you grow Manage the load Vs capacity Be clever with the redundancy strategy
Hard to do on a small scale? Predict the system capacity to get the end-game plan right – Room will generally end up too large running at too low a load with the future in mind Keep the UPS system load high without having too much modularity – Likely to be running at 30% load & 80% efficiency Introduce free-cooling – Most likely to be split air-con with CoP of 3-4 instead of
What could the impact be? 100 small data-rooms, 10 cabinets of 2kW each, 20kW IT load but designed for 50kW, each operating at PUE 3 of 2.5 – Total demand = 5MW One colo-facility of 200 cabinets of 6kW operating at PUE 3 of 1.3 – Total demand = 1.6MW – c70% reduction
Questions?
Next Break and Refreshments Next Presentation Steve Phipps