Network Management in Cloud and Edge Computing

✍ Scribed by Yuchao Zhang, Ke Xu

• Publisher: Springer
• Year: 2020
• Tongue: English
• Leaves: 148
• Edition: 1st ed. 2020
• Category: Library

✦ Synopsis

Traditional cloud computing and the emerging edge computing have greatly promoted the development of Internet applications. But what are the key issues in these two trends and what are the differences between them?

This book systematically introduces several key procedures in both cloud computing and edge computing scenarios, with each chapter providing a detailed description of novel design. In addition, the book also discusses a series of important findings from industry collaborations, which greatly enhance our understanding of the real system of industry. This book is not only a valuable reference resource for researchers, but also provides large-scale deployment cases for real systems in industry.

In order to gain the most benefit from this book, readers should have some the basic knowledge of computer networks.

✦ Table of Contents

Preface
Book Organization
Acknowledgments
Contents
1 Introduction
1.1 Research Background
1.2 Content Summary
1.3 Key Contributions
1.4 Chapter Arrangement
References
2 A Survey of Resource Management in Cloud and Edge Computing
2.1 Latency Optimization for Cloud-Based Service Chains
2.2 Toward Shorter Task Completion Time
2.3 Container Placement and Reassignment for Large-Scale Network
2.4 Near-Optimal Network System for Data Replication
2.5 Distributed Edge Caching in Short Video Network
2.6 The Controllability of Dynamic Temporal Network
References
3 A Task Scheduling Scheme in the DC Access Network
3.1 Introduction
3.2 Dynamic Differentiated Service with Delay-Guarantee
3.2.1 The Components of Latency
3.2.2 Design Philosophy
3.2.3 D3G Framework
3.2.4 Adjusting Resource Allocation
3.3 Deployment
3.4 D3G Experiment
3.4.1 Overall Performance
3.4.2 Algorithm Dynamism
3.4.3 System Scalability
3.5 Conclusion
References
4 A Cross-Layer Transport Protocol Design in the Terminal Systems of DC
4.1 Introduction
4.2 TAFA's Control Scheme
4.3 Key Challenges
4.4 TAFA: Task-Aware and Flow-Aware
4.4.1 Task-Awareness
4.4.1.1 End-Host Operations
4.4.1.2 Switch Operations
4.4.1.3 Multiple Priority Queues
4.4.2 Flow-Awareness
4.4.3 Algorithm Implementation
4.5 System Stability
4.6 TAFA Experiment
4.6.1 Setup
4.6.2 Overall Performance of TAFA
4.6.3 TAFA vs. Task-Aware
4.6.4 TAFA vs. Flow-Aware
4.7 Conclusion
References
5 Optimization of Container Communication in DC Back-EndServers
5.1 Container Group-Based Architecture
5.2 Problem Definition
5.2.1 Objective
5.2.1.1 Communication Cost
5.2.1.2 Resource Utilization Cost
5.2.1.3 Residual Resource Balance Cost
5.2.2 Constraints
5.3 Container Placement Problem
5.3.1 Problem Analysis
5.3.2 CA-WFD Algorithm
5.4 Container Reassignment Problem
5.4.1 Problem Analysis
5.4.2 Sweep&Search Algorithm
5.4.2.1 Sweep
5.4.2.2 Search
5.5 Implementation
5.6 Experiment
5.6.1 Performance of CA-WFD
5.6.1.1 Algorithm Performance
5.6.1.2 Algorithm Variations
5.6.2 Performance of Sweep&Search
5.6.2.1 Algorithm Performance
5.6.2.2 Algorithm Efficiency
5.7 Approximation Analysis of Sweep&Search
5.8 Conclusion
References
6 The Deployment of Large-Scale Data Synchronization System for Cross-DC Networks
6.1 Motivation of BDS+ Design
6.1.1 Baidu's Inter-DC Multicast Workload
6.1.2 Potentials of Inter-DC Application-Level Overlay
6.1.3 Limitations of Existing Solutions
6.1.4 Key Observations
6.2 System Overview
6.3 Near-Optimal Application-Level Overlay Network
6.3.1 Basic Formulation
6.3.2 Decoupling Scheduling and Routing
6.3.3 Scheduling
6.3.4 Routing
6.4 Dynamic Bandwidth Separation
6.4.1 Design Logic
6.4.2 Integrated to BDS+
6.4.2.1 Online Traffic Prediction Algorithm
6.4.2.2 Dynamic Bandwidth Separation
6.5 System Design
6.5.1 Centralized Control of BDS+
6.5.2 Dynamic Bandwidth Separation of BDS+
6.5.3 Fault Tolerance
6.5.4 Implementation and Deployment
6.6 BDS+ Experiment
6.6.1 BDS+ Over Existing Solutions
6.6.1.1 Methodology
6.6.1.2 BDS+ vs. Gingko
6.6.1.3 BDS+ vs. Other Overlay Multicast Techniques
6.6.2 Micro-benchmarks
6.6.2.1 Scalability
6.6.2.2 Fault Tolerance
6.6.2.3 Choosing the Values of Key Parameters
6.6.2.4 In-Depth Analysis
6.6.3 BDS+'s Dynamic Bandwidth Separation
6.6.3.1 Further Improvements Over BDS+
6.6.3.2 BDS+'s Prediction Algorithm
6.7 Conclusion
Appendix
References
7 Storage Issues in the Edge
7.1 Introduction
7.1.1 The Characteristics of Edge Caching in Short VideoNetwork
7.1.2 Limitations of Existing Solutions
7.2 AutoSight Design
7.2.1 System Overview
7.2.2 Correlation-Based Predictor: CoStore
7.2.3 Caching Engine: Viewfinder
7.3 AutoSight Experiment
7.3.1 Experiment Setting
7.3.2 Performance Comparison
7.4 Conclusion
References
8 Computing Issues in the Edge
8.1 Background
8.2 DND: Driver Node Algorithm
8.2.1 Parameters and Variable Declarations
8.2.2 Modeling
8.2.3 Abstraction of Topology
8.3 DND Experiment
8.3.1 Communication Radius
8.3.2 Nodes Density
8.3.3 Nodes Velocity
8.3.4 Control Time
8.4 Conclusion
References