Distributed Systems and Applications
Paraskevi Fragopoulou & Miltos Grammatikakis & Harris Papadakis
The course’s primary goal is to provide students with basic concepts and principles of large-scale dynamic distributed systems and distributed algorithms. The basic course material will include the following: after a general introduction to the topic of large scale distributed systems, a study of the most important architectures and technologies involved in modern distributed systems (i.e. peer-to-peer systems and cloud infrastructures). The course will lead to deepened knowledge in distributed systems and algorithms especially for fully decentralized peer-to-peer (P2P) systems, such as: structured overlay networks, gossip based systems and content distribution, computational cloud MapReduce and Hadoop.
The research orientation of the course will include the presentation of recent research publications in important conferences in the field. In the study and the presentation of these research papers the students following the course will be actively involved.
LessThe course’s primary goal is to provide students with basic concepts and principles of large-scale dynamic distributed systems and distributed algorithms. The basic course material will include the following: after a general introduction to the topic of large scale distributed systems, a study of the most important architectures and technologies involved in modern distributed systems (i.e. peer-to-peer systems and cloud infrastructures). The course will lead to deepened knowledge in distributed systems and algorithms especially for fully decentralized peer-to-peer (P2P) systems, such as: structured overlay networks, gossip based systems and content distribution, computational cloud MapReduce and Hadoop.
The research orientation of the course will include the presentation of recent research publications in important conferences in the field. In the study and the presentation of these research papers the students following the course will be actively involved.
The course’s primary goal is to provide students with basic concepts and principles of large-scale dynamic distributed systems and distributed algorithms. The basic course material will include the following: after a general introduction to the topic of large scale distributed systems, a study of the most important architectures and technologies involved in modern distributed systems (i.e. peer-to-peer systems and cloud infrastructures). The course will lead to deepened knowledge in distributed systems and algorithms especially for fully decentralized peer-to-peer (P2P) systems, such as: structured overlay networks, gossip based systems and content distribution, computational cloud MapReduce and Hadoop.
The research orientation of the course will include the presentation of recent research publications in important conferences in the field. In the study and the presentation of these research papers the students following the course will be actively involved.
-INTRODUCTION
-RELATED WORK
-SYSTEM MODEL
-THE CHORD PROTOCOL
-SIMULATION RESULTS
-FUTURE WORK
-CONCLUSION
-SYSTEM MODEL
-GOSSIP-BASED AGGREGATION
-The Basic Aggregation Protocol
-Theoretical Analysis of Gossip-Based Aggregation
-PRACTICAL PROTOCOL FOR GOSSIP-BASED AGGREGATION
-Automatic Restarting
-Coping with Churn
-THEORETICAL RESULTS FOR BENIGN FAILURES
-EXPERIMENTAL RESULTS ON PLANETLAB
-The problem
-The proposed solution
-The protocol
-Optimizations
-Simulation Experiments
-Self-healing
-Age-based view update
-Experimental results
-Application Examples
-Clustering and Sorting
-Overview of BitTorrent
-Tracker protocol
-Peer wire protocol
-New features
-DHT (Distributed Hash Table)
-Multi trackers
-Encryption
-P2P media streaming
-Related Solutions in Media Streaming
-Locating supplying peers
-Centralized directory
-Hierarchical overlay structure
-DHT-based approach
-Other approaches and comparisons
-Content delivery path maintenance
-Tree-based multicast
-Multiple trees
-Pull-based gossiping
-Live-Streaming on the Gradient Overlay Network
-RELATED WORK
-PROBLEM DESCRIPTION
-SEPIDAR SYSTEM
-Gradient overlay construction
-Streaming tree overlay construction
-Freerider detection and punishment
-EXPERIMENTS AND EVALUATION
-Establishing parameters for good system performance
-Freerider detector settings
-Sepidar vs. NewCoolstreaming
-Incentivizing nodes to contribute upload bandwidth
-Programming Model
-More Examples
-Implementation
-Execution Overview
-Master Data Structures
-Fault Tolerance
-Worker Failure
-Semantics in the Presence of Failures
-Task Granularity
-Backup Tasks
-Ordering Guarantees
-Input and Output Types
-Status Information
-Cluster Con[1]guration
-Background
-Hadoop Architecture
-Programming Model
-Map Task Execution
-Reduce Task Execution
-Pipelined MapReduce
-PipeliningWithin A Job
-Fault Tolerance
-Pipelining Between Jobs
-Task Scheduling
-Online Aggregation
-Single-Job Online Aggregation
-Multi-Job Online Aggregation
-Continuous Queries
-Continuous MapReduce Jobs
-Prototype Monitoring System
-Performance Evaluation
-Executive Summary
-Cloud Computing: An Old Idea Whose Time Has (Finally) Come
-What is Cloud Computing?
-Clouds in a Perfect Storm: Why Now, Not Then?
-New Application Opportunities
-Classes of Utility Computing
-Cloud Computing Economics
-Comparing Costs: Should I Move to the Cloud?
-Top 10 Obstacles and Opportunities for Cloud Computing
-Number 2 Obstacle: Data Lock-In
-Number 3 Obstacle: Data Confidentiality and Auditability
-Conclusion and Questions about the Clouds of Tomorrow
Open Academic Course
Num. of Visits : 336
Num. of Hits : 9984
