RELATED BACKGROUND WORK OZLEM BILGIR. OUTLINE 1- Gandhi et al., Optimal Power Allocation in Server Farms, SIGMETRICS’09 2-Chen et al., Managing Server.

Slides:



Advertisements
Similar presentations
Exploring the Potential of CMP Core Count Management on Data Center Energy Savings Ozlem Bilgir * Margaret Martonosi * Qiang Wu * Princeton University.
Advertisements

Dynamic Power Redistribution in Failure-Prone CMPs Paula Petrica, Jonathan A. Winter * and David H. Albonesi Cornell University *Google, Inc.
Gregory Shklover, Ben Emanuel Intel Corporation MATAM, Haifa 31015, Israel Simultaneous Clock and Data Gate Sizing Algorithm with Common Global Objective.
Chapter 7 1 Cellular Telecommunications Systems Abdulaziz Mohammed Al-Yami
© 2013 The SmartenIT Consortium 1 Commercial in Confidence Game Theoretic approach to energy efficiency Mateusz Wielgosz, Krzysztof Wajda, AGH Krakow Meeting,
1 Distributed Adaptive Sampling, Forwarding, and Routing Algorithms for Wireless Visual Sensor Networks Johnsen Kho, Long Tran-Thanh, Alex Rogers, Nicholas.
Silberschatz, Galvin and Gagne ©2009 Operating System Concepts – 8 th Edition Chapter 5: CPU Scheduling.
A Cyber-Physical Systems Approach to Energy Management in Data Centers Presented by Chen He Adopted form the paper authors.
CSE 691: Energy-Efficient Computing Lecture 4 SCALING: stateless vs. stateful Anshul Gandhi 1307, CS building
Aleksandar Kuzmanovic and Edward W. Knightly Rice Networks Group Measuring Service in Multi-Class Networks.
Energy Efficient Dynamic Provisioning in Data Centers: The Benefit of Seeing the Future TexPoint fonts used in EMF. Read the TexPoint manual before you.
Helmholtz International Center for Oliver Boine-Frankenheim GSI mbH and TU Darmstadt/TEMF FAIR accelerator theory (FAIR-AT) division Helmholtz International.
Cost Tradeoff of Consistency Over Data Centers Ozlem Bilgir.
SIGMETRICS 2008: Introduction to Control Theory. Abdelzaher, Diao, Hellerstein, Lu, and Zhu. CPU Utilization Control in Distributed Real-Time Systems Chenyang.
Energy Management and Adaptive Behavior Tarek Abdelzaher.
Scheduling in Batch Systems
Selfish Caching in Distributed Systems: A Game-Theoretic Analysis By Byung-Gon Chun et al. UC Berkeley PODC’04.
An Experimental Evaluation on Reliability Features of N-Version Programming Xia Cai, Michael R. Lyu and Mladen A. Vouk ISSRE’2005.
Chapter 5-CPU Scheduling
U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science Virtualization in Data Centers Prashant Shenoy
Proteus: Power Proportional Memory Cache Cluster in Data Centers Shen Li, Shiguang Wang, Fan Yang, Shaohan Hu, Fatemeh Saremi, Tarek Abdelzaher.
A Multi-Agent Learning Approach to Online Distributed Resource Allocation Chongjie Zhang Victor Lesser Prashant Shenoy Computer Science Department University.
Power saving technique for multi-hop ad hoc wireless networks.
Power Management in Data Centers: Theory & Practice Mor Harchol-Balter Computer Science Dept Carnegie Mellon University 1 Anshul Gandhi, Sherwin Doroudi,
Project Management and Scheduling
Power Management in Data Centers: Theory & Practice Mor Harchol-Balter Computer Science Dept Carnegie Mellon University 1 Anshul Gandhi, Sherwin Doroudi,
Computer Science Cataclysm: Policing Extreme Overloads in Internet Applications Bhuvan Urgaonkar and Prashant Shenoy University of Massachusetts.
Power Containers: An OS Facility for Fine-Grained Power and Energy Management on Multicore Servers Kai Shen, Arrvindh Shriraman, Sandhya Dwarkadas, Xiao.
A User Experience-based Cloud Service Redeployment Mechanism KANG Yu.
Flexible Channelization for Wireless LANs Zafar Ayyub Qazi*, Zhibin Dou and Prof. Samir Das* *Department of Computer Science (WINGS lab), Stony.
CONGRESSIONAL SAMPLES FOR APPROXIMATE ANSWERING OF GROUP-BY QUERIES Swarup Acharya Phillip Gibbons Viswanath Poosala ( Information Sciences Research Center,
Database Replication Policies for Dynamic Content Applications Gokul Soundararajan, Cristiana Amza, Ashvin Goel University of Toronto EuroSys 2006: Leuven,
EmNet: Satisfying The Individual User Through Empathic Home Networks J. Scott Miller, John R. Lange & Peter A. Dinda Department of Electrical Engineering.
1 Heterogeneity in Multi-Hop Wireless Networks Nitin H. Vaidya University of Illinois at Urbana-Champaign © 2003 Vaidya.
Glasgow 02/02/04 NN k networks for content-based image retrieval Daniel Heesch.
DISCERN: Cooperative Whitespace Scanning in Practical Environments Tarun Bansal, Bo Chen and Prasun Sinha Ohio State Univeristy.
Silberschatz and Galvin  Operating System Concepts Module 5: CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms Multiple-Processor.
Introduction The berthing assignment problem requires that a detailed time-and-space-schedule be planned for incoming ships, with the goal of minimizing.
Optimization of PHEV/EV Battery Charging Lawrence Wang CURENT YSP Presentations RM :00-11:25 1.
A dynamic optimization model for power and performance management of virtualized clusters Vinicius Petrucci, Orlando Loques Univ. Federal Fluminense Niteroi,
Towards Dynamic Green-Sizing for Database Servers Mustafa Korkmaz, Alexey Karyakin, Martin Karsten, Kenneth Salem University of Waterloo.
Managing Server Energy and Operational Costs Chen, Das, Qin, Sivasubramaniam, Wang, Gautam (Penn State) Sigmetrics 2005.
Slide 1 Tutorial: Optimal Learning in the Laboratory Sciences Forming the decision set December 10, 2014 Warren B. Powell Kris Reyes Si Chen Princeton.
@ Carnegie Mellon Databases 1 Finding Frequent Items in Distributed Data Streams Amit Manjhi V. Shkapenyuk, K. Dhamdhere, C. Olston Carnegie Mellon University.
Adapting Channel Widths to Improve Application Performance Ranveer Chandra Microsoft Research Collaborators: Victor Bahl, Ratul Mahajan, Thomas Moscibroda,
CSE 591: Energy-Efficient Computing Lecture 3 SPEED: processor Anshul Gandhi 347, CS building
Basic Concepts Maximum CPU utilization obtained with multiprogramming
Dynamic Resource Allocation for Shared Data Centers Using Online Measurements By- Abhishek Chandra, Weibo Gong and Prashant Shenoy.
Module 6: Configuring and Managing Windows SharePoint Services 3.0.
DASH2M: Exploring HTTP/2 for Internet Streaming to Mobile Devices
Data Driven Resource Allocation for Distributed Learning
Dan C. Marinescu Office: HEC 439 B. Office hours: M, Wd 3 – 4:30 PM.
Trading Timeliness and Accuracy in Geo-Distributed Streaming Analytics
Measuring Service in Multi-Class Networks
Comparison of the Three CPU Schedulers in Xen
Columbia University in the city of New York
Optimization of PHEV/EV Battery Charging
Module 5: CPU Scheduling
Chapter 5: CPU Scheduling
Apollo Weize Sun Feb.17th, 2017.
Chapter 2 Basic Models for the Location Problem
TimeTrader: Exploiting Latency Tail to Save Datacenter Energy for Online Search Balajee Vamanan, Hamza Bin Sohail, Jahangir Hasan, and T. N. Vijaykumar.
3: CPU Scheduling Basic Concepts Scheduling Criteria
Chapter5: CPU Scheduling
Chapter 5: CPU Scheduling
Chapter 5: CPU Scheduling
Joint Processing MU-MIMO
Module 5: CPU Scheduling
Module 5: CPU Scheduling
Business Planning Budgeting
Presentation transcript:

RELATED BACKGROUND WORK OZLEM BILGIR

OUTLINE 1- Gandhi et al., Optimal Power Allocation in Server Farms, SIGMETRICS’09 2-Chen et al., Managing Server Energy and Operational Costs in Hosting Centers, SIGMETRICS’05 3- Raghavendra et al., No “Power” Struggles: Coordinated Multi-level Power Management for the Data Center, ASPLOS’08

Gandhi et al., SIGMETRICS’09 What is the best power allocation model for better response time? Less servers with higher frequency? More servers with lower frequency?

Gandhi et al., SIGMETRICS’09 (cont.) Optimal power allocation depends on “power-frequency relationship” Power-frequency relationship depends on many factors.

Gandhi et al., SIGMETRICS’09 (cont.) Power- frequency relation depends on: Voltage/frequency scaling technique

Gandhi et al., SIGMETRICS’09 (cont.) Experimental Results: Open Server Farm;

Gandhi et al., SIGMETRICS’09 (cont.) Experimental Results: Close Server Farm;

Gandhi et al., SIGMETRICS’09 (cont.) Summary: -Many factors( frequency scaling technique, arrival rate, total power budget, open/close server configuration.. etc.) affect mean response time

Chen et al., SIGMETRICS’05 Data Centers should be able to meet SLA Over-provisioning service capacity to meet SLA for worst case is one option Urgaonkar et. Al, Ranjan et. Al. Chandra et.al looked at finding right capacity and distributing right capacity across different applications to meet SLA Still wasting servers, hence money!!!

Chen et al., SIGMETRICS’05(cont.) m1 m2 m3

Chen et al., SIGMETRICS’05(cont.) Problem Formulation & Methodology: Total Cost: Server allocation & f modulation period

Chen et al., SIGMETRICS’05(cont.) 3 different approaches; – Queuing Theory Based Approach – Control Theoretic Approach – Hybrid Approach

Chen et al., SIGMETRICS’05(cont.) 3 different approaches; – Queuing Theory Based Approach – Control Theoretic Approach – Hybrid Approach

Chen et al., SIGMETRICS’05(cont.) Queuing Theory Based Approach: where

Chen et al., SIGMETRICS’05(cont.) Control Theoretic Approach:

Chen et al., SIGMETRICS’05(cont.) Feedback Control of Aggregate Frequency Rf & Rwi are the weights

Chen et al., SIGMETRICS’05(cont.) Server Allocation: # of servers for min. Energy:

Chen et al., SIGMETRICS’05(cont.) Server allocation Decision Mechanism: 1. m(u-1) > min(m*(u), m(u)) m(u) = min(m*(u), m(u)) 2. otherwise; Cost of turning on one server: Bo Red. in power cost:

Chen et al., SIGMETRICS’05(cont.) Summary: -Control theoretical method is similar, but not the same with my works -Turning on-off/DVFS decision mechanism can be improved

Raghavendra et al., ASPLOS’08 There are many separate power management solutions addressing different problems Their collection may be non-optimal or unstable There is a need for coordinated solution which is a combination of separate solutions

Raghavendra et al., ASPLOS’08(cont.)

Summary: -Combining different techniques will give better, more stable, and consistent results