Deadline Scheduling and Heavy tailED distributions

Slides:

Advertisements

Similar presentations

1 Real-Time Queueing Theory Presented by: John Lehoczky Carnegie Mellon SAMSI Workshop Congestion Control and Heavy Traffic.

Advertisements

VARUN GUPTA Carnegie Mellon University 1 With: Mor Harchol-Balter (CMU)

To Queue or Not to Queue? Physical queues can be really stressful and exhausting…

State-Space Collapse via Drift Conditions Atilla Eryilmaz (OSU) and R. Srikant (Illinois) 4/10/20151.

Speaker: Yu-Fu Huang Advisor: Dr. Kai-Wei Ke Date : 2014, Mar. 17 A page-oriented WWW traffic model for wireless system simulations.

Hadi Goudarzi and Massoud Pedram

System-level Trade-off of Networks-on-Chip Architecture Choices Network-on-Chip System-on-Chip Group, CSE-IMM, DTU.

An Analytical Model for Network Flow Analysis Ernesto Gomez, Yasha Karant, Keith Schubert Institute for Applied Supercomputing Department of Computer Science.

Network Design and Optimization Queueing Theory Overview Dr. Greg Bernstein Grotto Networking

Crew: Pavel Babenkov Kirill Kalinkin Customized Stochastic Lot Scheduling Problem with Due Date Management.

Load Balancing of Elastic Traffic in Heterogeneous Wireless Networks Abdulfetah Khalid, Samuli Aalto and Pasi Lassila

Competitive Analysis of Buffer Policies with SLA Commitments Boaz Patt-Shamir, Tel Aviv University Gabriel Scalosub, University of Toronto Yuval Shavitt,

Managing Risk of Inaccurate Runtime Estimates for Deadline Constrained Job Admission Control in Clusters Chee Shin Yeo and Rajkumar Buyya Grid Computing.

IFIP Performance 2007 On Processor Sharing (PS) and Its Applications to Cellular Data Network Provisioning Yujing Wu, Carey Williamson, Jingxiang Luo Department.

Towards Feasibility Region Calculus: An End-to-end Schedulability Analysis of Real- Time Multistage Execution William Hawkins and Tarek Abdelzaher Presented.

Analyzing Multi-channel MAC Protocols for Underwater Sensor Networks Presenter: Zhong Zhou.

1 HEURISTICS FOR DYNAMIC SCHEDULING OF MULTI-CLASS BASE-STOCK CONTROLLED SYSTEMS Bora KAT and Zeynep Müge AVŞAR Department of Industrial Engineering Middle.

CAC for Multimedia Services in Mobile Cellular Networks ： A Markov Decision Approach Speaker ： Xu Jia-Hao Advisor ： Ke Kai-Wei Date ： 2004 / 11 / 18.

Parallel Job Scheduling Algorithms and Interfaces Research Exam for Cynthia Bailey Lee Department of Computer Science and Engineering University of California,

Optimization for QoS on Systems with Tasks Deadlines Luis Fernando Orleans Pedro Nuno Furtado.

Björn Landfeldt School of Information Technologies Investigating a theoretical model Bjorn Landfeldt University of Sydney.

Performance Analysis of Wavelength-Routed Optical Networks with Connection Request Retrials Fei Xue+, S. J. Ben Yoo+, Hiroyuki Yokoyama*, and Yukio Horiuchi*

A Hierarchical Characterization of a Live Streaming Media Workload IEEE/ACM Trans. Networking, Feb Eveline Veloso, Virg í lio Almeida, Wagner Meira,

I think your suggestion is, Can we do two things at once? Well, we’re of the view that we can walk and chew gum at the same time. —Richard Armitage, deputy.

Effect of higher moments of job size distribution on the performance of an M/G/k system VARUN GUPTA Joint work with: Mor Harchol-Balter Carnegie Mellon.

Effect of higher moments of job size distribution on the performance of an M/G/k system VARUN GUPTA Joint work with: Mor Harchol-Balter Carnegie Mellon.

Dimitrios Konstantas, Evangelos Grigoroudis, Vassilis S. Kouikoglou and Stratos Ioannidis Department of Production Engineering and Management Technical.

1 Real-Time Queueing Network Theory Presented by Akramul Azim Department of Electrical and Computer Engineering University of Waterloo, Canada John P.

References for M/G/1 Input Process

Integrated Risk Analysis for a Commercial Computing Service Chee Shin Yeo and Rajkumar Buyya Grid Computing and Distributed Systems (GRIDS) Lab. Dept.

Multiple-access Communication in Networks A Geometric View W. Chen & S. Meyn Dept ECE & CSL University of Illinois.

OPTIMAL SERVER PROVISIONING AND FREQUENCY ADJUSTMENT IN SERVER CLUSTERS Presented by: Xinying Zheng 09/13/ XINYING ZHENG, YU CAI MICHIGAN TECHNOLOGICAL.

Exploiting Clustering Techniques for Web Session Inference A.Bianco, G. Mardente, M. Mellia, M.Munafò, L. Muscariello (Politecnico di Torino)

© 2009 IBM Corporation 1 Improving Consolidation of Virtual Machines with Risk-aware Bandwidth Oversubscription in Compute Clouds Amir Epstein Joint work.

Scheduling Periodic Real-Time Tasks with Heterogeneous Reward Requirements I-Hong Hou and P.R. Kumar 1.

1 A Simple Asymptotically Optimal Energy Allocation and Routing Scheme in Rechargeable Sensor Networks Shengbo Chen, Prasun Sinha, Ness Shroff, Changhee.

CS433 Modeling and Simulation Lecture 12 Queueing Theory Dr. Anis Koubâa 03 May 2008 Al-Imam Mohammad Ibn Saud University.

Tactical Planning in Healthcare with Approximate Dynamic Programming Martijn Mes & Peter Hulshof Department of Industrial Engineering and Business Information.

Scheduling Periodic Real-Time Tasks with Heterogeneous Reward Requirements I-Hong Hou and P.R. Kumar 1 Presenter: Qixin Wang.

Opportunistic Traffic Scheduling Over Multiple Network Path Coskun Cetinkaya and Edward Knightly.

An Optimal Distributed Call Admission control for Adaptive Multimedia in Wireless/Mobile Networks Reporter: 電機所鄭志川.

Maximizing Lifetime per Unit Cost in Wireless Sensor Networks

1/22 Optimization of Google Cloud Task Processing with Checkpoint-Restart Mechanism Speaker: Sheng Di Coauthors: Yves Robert, Frédéric Vivien, Derrick.

Analysis of RED Goal: impact of RED on loss and delay of bursty (TCP) and less bursty or smooth (UDP) traffic RED eliminates loss bias against bursty traffic.

OPERATING SYSTEMS CS 3530 Summer 2014 Systems and Models Chapter 03.

Risk Analysis Workshop April 14, 2004 HT, LRD and MF in teletraffic1 Heavy tails, long memory and multifractals in teletraffic modelling István Maricza.

Project Background My project goal was to accurately model a dipole in the presence of the lossy Earth. I used exact image theory developed previously.

Xi He Golisano College of Computing and Information Sciences Rochester Institute of Technology Rochester, NY THERMAL-AWARE RESOURCE.

Courtesy Piggybacking: Supporting Differentiated Services in Multihop Mobile Ad Hoc Networks Wei LiuXiang Chen Yuguang Fang WING Dept. of ECE University.

THE DEADLINE-BASED SCHEDULING OF DIVISIBLE REAL-TIME WORKLOADS ON MULTIPROCESSOR PLATFORMS Suriayati Chuprat Supervisors: Professor Dr Shaharuddin Salleh.

IBM Thomas J. Watson Research Center © 2004 IBM Corporation Analysis of Parallel-Server Systems with Dynamic Affinity Scheduling and Load Balancing Mark.

OPERATING SYSTEMS CS 3502 Fall 2017

Simulation and Exploration of

An Analytical Model for Network Flow Analysis

Load Balancing and Data centers

Serve Assignment Policies

Al-Imam Mohammad Ibn Saud University

Dynamic Graph Partitioning Algorithm

Deadline Scheduling and Heavy tail distributionS

Queuing Theory Non-Markov Systems

Measuring Service in Multi-Class Networks

CPSC 531: System Modeling and Simulation

Moab® Automation Intelligence Overview

 Real-Time Scheduling via Reinforcement Learning

Web switch support for differentiated services

Network Traffic Modeling

 Real-Time Scheduling via Reinforcement Learning

Cloud Computing and Multivariate Heavy Tails

Schedulability Conditions for Scheduling Algorithms

Resource Sharing with Subexponential Distributions

Presentation transcript:

Deadline Scheduling and Heavy tailED distributions Lang Tong School of Electrical and Computer Engineering Cornell University, Ithaca, NY 14850

Overview Objective Research problem and applications Collaborations Develop design principles and tools for job scheduling problems with multivariate stochastic characteristics in a distributed and hierarchical computation system. Characterize impact of heavy tailed distribution on system performance Research problem and applications Deadline scheduling with outsourcing options Tactical cloud computing for battlefield situation awareness Collaborations N. Shroff, R. Srikant. Ananthram Swami (ARL) Yu Zhe and Shiyao Chen

Deadline scheduling and outsourcing Key characteristics: Jobs arrive randomly with different sizes, deadlines, and rewards Multiple local inexpensive servers Option of outsourcing to meet workload requirement and quality of service Multivariate distributions: job size, inter-arrival, deadline. Scheduling action space: Admission: whether to take on a job Scheduling: which job to process and on which server Outsourcing: whether and when outsource

Outline Problem formulation Scheduling policy and optimality Structure of optimal policy TAGS (Threshold admission and greedy scheduling) Stochastic TAGS and optimality Performance analysis An upper bound and a heavy traffic approximation Numerical results and insights Conclusions and future work

Jobs with deadlines

Scheduling actions

Performance measure:

Optimality: post admission scheduling

Optimality: competitive ratio optimality

Stochastic TAGS

Performance examples Poisson arrival, Identical job length, Identical laxity Pareto arrival, Pareto job length, Pareto laxity, within light tail region (finite variance)

Outline Problem formulation Scheduling policy and optimality Structure of optimal policy TAGS (Threshold admission and greedy scheduling) Stochastic TAGS and optimality Performance analysis An upper bound and a heavy traffic approximation of upper bound. Numerical results and insights Conclusions and future work

Performance upper bound

Heavy traffic approximation

Performance evaluation

Performance examples: (easy jobs) Poisson arrival, Pareto job length, Pareto laxity (EPP, alpha=4)

Performance examples: (easy jobs) Pareto arrival, Pareto job length, Pareto laxity (PPP, alpha=4)

Performance examples: (tight jobs) Pareto arrival, Pareto job length, Exponential laxity (PPE, alpha=4)

Clustered arrivals and tight deadlines Pareto arrival (2), Pareto job length (2), Pareto laxity (0.5) (alpha=4)

Effects of tail parameters Pareto arrival, Pareto job length, Pareto laxity (PPP)

Summary and future work Deadline scheduling with outsourcing options TAGS (Threshold admission and greedy scheduling) Simple online scheduling Optimality Performance A performance upper bound and approximations Simulation results: accurate for jobs with low reneging fraction. inaccurate approximation for tight jobs. Current and future work Multi-processor deadline scheduling problems Large system asymptotic properties