A Computer Scientist Looks at ( Energy) Sustainability

1. A Computer ScientistLooks at(Energy) Sustainability Randy H. Katz University of California, Berkeley NSF Sustainable Energy-EfficientData Management Workshop 2 May 2011

2. Working Definition • Sustainability: strategies that meet society’s present needs without compromising the ability of future generations to meet their own needs • Satisfaction of basic economic, social, and security needs now and in the future without undermining the natural resource base and environmental quality on which life depends • Energy, Water, Natural Resources, Environment, … • Green Manufacturing, Transportation, Agriculture, ... • Increase efficiencies and minimize bad side-effects • Use less, use what you need, use better

3. Role ofInformation Technology • “Energy permits things to exist; information, to behave purposefully.” W. Ware, 1997 • Observe-Analyze-Act • Observe: Sense, Monitor, Collect • Analyze: Organize, Model, Infer, Plan • Act: Control, Actuate, Execute, Manage

4. Energy “Spaghetti” Chart Quads (1015 BTUs) http://www.eia.doe.gov 2008 data 4

5. Sources and Loads Dispatchable Sources Non-Dispatchable Sources Oblivious Loads Aware Loads

6. Supply- versus Load- Following Load Duration Curve Most expensive, least efficient energy Latency involved in bringing capacity on line PeakerCapacity Load-followingSupply IntermediateCapacity Demand Response: Incentivize reduced loads duringtimes of peak demand Demand Side Management: Shift demand to reduce peak loads, e.g., Supply-following Loads Base Capacity (or probability of exceeding)

7. 21st Century Energy Infrastructure • Energy: the limited resource of the 21st Century • Role of IT: Information Age approach • Overarching Principle: bits follow current • Pervasive actionable awareness of energy availability and consumption, on fine time scales • Exploit information to match sources to loads, manage buffers, integrate renewables, signal demand response, and take advantage of locality

8. Energy Networks Gen-to-Building Gen-to-Grid Facility-to-Building Building-to-Grid Building-to-Grid Facility-to-Building uGrid-to-Grid Facility-to-Building Grid Storage-to-Building Building MR-to-Building Demand Response Machine Room Load Following Temperature Maintenance Supply Following Plug Loads Instrumentation Models Web Server Grid OS Instrumentation Instrumentation Lighting CompressorScheduling Web App Logic Instrumentation Models Models Models DB/Storage Facilities Routing/Control Control Building OS Load Balancer/Scheduler Supply-FollowingLoads Power-AwareCluster Manager Controls

9. Datacenter as aSupply-Following Load • Degree of Freedom: On-demand + scheduled workloads • Principle: Power proportionality from non-power proportional components • Sustainability: Maximize use of renewable sources

10. Supply-side Challenge:Wind Single Location Multiple Locations vs. Time of Year • High variability of wind energy is an impediment to its large-scale penetration in traditional Grid/Load architectures Std dev / mean +/-11MW in 20 min

11. Load-side Challenge:Power Proportionality Work Capacity Static Load Provisioning: 100% overprovisioning over worst expected case • Scheduling agility: workload awareness and resource allocation • Wikipedia interactive workload + HPC batch workload Over- provisioning or Spinning Reserve Dynamic Capacity Management Dynamic Load Arrival Time

12. Energy-AwareSystem Architecture Internet Load Balancer / Job Scheduler Database / SAN Clients AWAKE Web App Web App TRANSITION Web App Web App Web Server App Server Web Server Web Server Batch Server SLEEPING Add/Remove Nodes Delay/Degrade Jobs Workload/Request Statistics Utilization, Response Times,Dropped Requests Wake/Sleep Cluster Manager Price/ Renewable Energy Load Monitor Workload Prediction Cluster Provisioning Energy Policy Grid Electrical Power Battery Charge/Discharge Energy Usage Power Subsystem: Metering, Distribution, Battery Control Observe Analyze Act

13. Server Efficiencies Server Class Machines (similar figure for netbook/embedded class nodes) Better Better Operating Range Measured

14. Effectively Scaling Work Capacity and Power Measured experiment on LoClu

15. Information Overlay to the Energy Grid Intelligent Energy Network Source IPS energy subnet Load IPS Intelligent Power Switch Generation Transmission Distribution Load Conventional Electric Grid Conventional Internet

16. Aware Co-operative Grid Power Proportional Cluster as a Model Systemapplied to the Smart Grid—now distributed • Availability • Pricing • Planning • Forecasting • Tracking • Market • Observe-Analyze-Act: • Deep instrumentation • Waste elimination • Efficient Operation • Shifting, Scheduling, Adaptation

17. Smart Buildings Cory Hall SDH Soda Hall

18. Smart Buildings External Human-BuildingInterface Control Plan and Schedule Building Integrated Operating System Activity/Usage Streams Occupant Demand Multi-Objective Model-Driven Control Pervasive Sensing Fault, Attack, Anomaly Detect &Management Activity Models Process Loads Physical Models PIB Electrical Transport BMS Legacy Instrumentation & Control Interfaces HVAC BITS BIM Light Cyber Physical Building Observe – Analyze – Act

19. Building Observation

20. Building Analysis Operational Environmental

21. BuildingAnalysis Soda Hall@Berkeley:IT-intensive CS Building Lighting HVAC IT and Plug Load PDUs, CRACs Servers

22. Building Action 50 Ton Chiller 200 Ton Chiller Etchevery Hall Soda Hall 2 months 10 months Scott McNally Bldg Manager

23. Building OS • sMAP: Simple Measurement and Actuation Profile for Physical Information • StreamFS: StorageSystem for energysensor data • Scheduling and Slack/Supply-following Loads

24. Summary • Awareness of Load and Supply • Load-Following: match load with managed supply • Demand Response: reduce load to meet supply • Supply-Following: schedule work to exploit knowledge of available supply—essential for non-dispatchable sources like wind and solar • Key idea: make information actionable • Observe-Analyze-Act • Information overlay on cluster, machine room, building-scale “grids” • Interface sensors, facilities, clusters, and buildings to information buses at a variety of scales

25. Conclusions • Smart Clusters, Smart Buildings, Smart Grids • Use less energy • Right provisioning for expected + reserve vs. peak • Use the energy you need: • Power proportionality • Use better energy • Integrate renewables

26. Thank You!