Service Quality of Cloud-Based Applications

ERIC BAUER, MS, is Reliability Engineering Manager in the IP Platforms CTO of Alcatel-Lucent. The holder of more than a dozen U.S. patents, Mr. Bauer is the author or coauthor of Reliability and Availability of Cloud Computing, Beyond Redundancy: How Geographic Redundancy Can Improve Service Availability and Reliability of Computer-Based Systems, Design for Reliability: Information and Computer-Based Systems, and Practical System Reliability.RANDEE ADAMS, MS, is a Consulting Member of the Technical Staff in the IP Platforms CTO of Alcatel-Lucent. She is the coauthor of Beyond Redundancy: How Geographic Redundancy Can Improve Service Availability and Reliability of Computer-Based Systems and Reliability and Availability of Cloud Computing. Ms. Adams has worked on many projects, including software development, delivery, engineering, architecture, and design.

• Figures xv Tables and Equations xxi1 INTRODUCTION 11.1 Approach 11.2 Target Audience 31.3 Organization 3I CONTEXT 72 APPLICATION SERVICE QUALITY 92.1 Simple Application Model 92.2 Service Boundaries 112.3 Key Quality and Performance Indicators 122.4 Key Application Characteristics 152.5 Application Service Quality Metrics 172.6 Technical Service versus Support Service 272.7 Security Considerations 283 CLOUD MODEL 293.1 Roles in Cloud Computing 303.2 Cloud Service Models 303.3 Cloud Essential Characteristics 313.4 Simplifi ed Cloud Architecture 333.5 Elasticity Measurements 363.6 Regions and Zones 443.7 Cloud Awareness 454 VIRTUALIZED INFRASTRUCTURE IMPAIRMENTS 494.1 Service Latency, Virtualization, and the Cloud 504.2 VM Failure 544.3 Nondelivery of Configured VM Capacity 544.4 Delivery of Degraded VM Capacity 574.5 Tail Latency 594.6 Clock Event Jitter 604.7 Clock Drift 614.8 Failed or Slow Allocation and Startup of VM Instance 624.9 Outlook for Virtualized Infrastructure Impairments 63II ANALYSIS 655 APPLICATION REDUNDANCY AND CLOUD COMPUTING 675.1 Failures, Availability, and Simplex Architectures 685.2 Improving Software Repair Times via Virtualization 705.3 Improving Infrastructure Repair Times via Virtualization 725.4 Redundancy and Recoverability 755.5 Sequential Redundancy and Concurrent Redundancy 805.6 Application Service Impact of Virtualization Impairments 845.7 Data Redundancy 905.8 Discussion 926 LOAD DISTRIBUTION AND BALANCING 976.1 Load Distribution Mechanisms 976.2 Load Distribution Strategies 996.3 Proxy Load Balancers 996.4 Nonproxy Load Distribution 1016.5 Hierarchy of Load Distribution 1026.6 Cloud-Based Load Balancing Challenges 1036.7 The Role of Load Balancing in Support of Redundancy 1036.8 Load Balancing and Availability Zones 1046.9 Workload Service Measurements 1046.10 Operational Considerations 1056.11 Load Balancing and Application Service Quality 1077 FAILURE CONTAINMENT 1117.1 Failure Containment 1117.2 Points of Failure 1167.3 Extreme Solution Coresidency 1227.4 Multitenancy and Solution Containers 1248 CAPACITY MANAGEMENT 1278.1 Workload Variations 1288.2 Traditional Capacity Management 1298.3 Traditional Overload Control 1298.4 Capacity Management and Virtualization 1318.5 Capacity Management in Cloud 1338.6 Storage Elasticity Considerations 1358.7 Elasticity and Overload 1368.8 Operational Considerations 1378.9 Workload Whipsaw 1388.10 General Elasticity Risks 1408.11 Elasticity Failure Scenarios 1419 RELEASE MANAGEMENT 1459.1 Terminology 1459.2 Traditional Software Upgrade Strategies 1469.3 Cloud-Enabled Software Upgrade Strategies 1539.4 Data Management 1589.5 Role of Service Orchestration in Software Upgrade 1599.6 Conclusion 16110 END-TO-END CONSIDERATIONS 16310.1 End-to-End Service Context 16310.2 Three-Layer End-to-End Service Model 16910.3 Distributed and Centralized Cloud Data Centers 17710.4 Multitiered Solution Architectures 18310.5 Disaster Recovery and Geographic Redundancy 184III RECOMMENDATIONS 19111 ACCOUNTABILITIES FOR SERVICE QUALITY 19311.1 Traditional Accountability 19311.2 The Cloud Service Delivery Path 19411.3 Cloud Accountability 19711.4 Accountability Case Studies 20011.5 Service Quality Gap Model 20511.6 Service Level Agreements 21012 SERVICE AVAILABILITY MEASUREMENT 21312.1 Parsimonious Service Measurements 21412.2 Traditional Service Availability Measurement 21512.3 Evolving Service Availability Measurements 21712.4 Evolving Hardware Reliability Measurement 22612.5 Evolving Elasticity Service Availability Measurements 22812.6 Evolving Release Management Service Availability Measurement 22912.7 Service Measurement Outlook 23113 APPLICATION SERVICE QUALITY REQUIREMENTS 23313.1 Service Availability Requirements 23413.2 Service Latency Requirements 23713.3 Service Reliability Requirements 23713.4 Service Accessibility Requirements 23813.5 Service Retainability Requirements 23913.6 Service Throughput Requirements 23913.7 Timestamp Accuracy Requirements 24013.8 Elasticity Requirements 24013.9 Release Management Requirements 24113.10 Disaster Recovery Requirements 24114 VIRTUALIZED INFRASTRUCTURE MEASUREMENT AND MANAGEMENT 24314.1 Business Context for Infrastructure Service Quality Measurements 24414.2 Cloud Consumer Measurement Options 24514.3 Impairment Measurement Strategies 24714.4 Managing Virtualized Infrastructure Impairments 25215 ANALYSIS OF CLOUD-BASED APPLICATIONS 25515.1 Reliability Block Diagrams and Side-by-Side Analysis 25615.2 IaaS Impairment Effects Analysis 25715.3 PaaS Failure Effects Analysis 25915.4 Workload Distribution Analysis 26015.5 Anti-Affi nity Analysis 26215.6 Elasticity Analysis 26315.7 Release Management Impact Effects Analysis 26715.8 Recovery Point Objective Analysis 26815.9 Recovery Time Objective Analysis 27016 TESTING CONSIDERATIONS 27316.1 Context for Testing 27316.2 Test Strategy 27416.3 Simulating Infrastructure Impairments 27716.4 Test Planning 27817 CONNECTING THE DOTS 28717.1 The Application Service Quality Challenge 28717.2 Redundancy and Robustness 28917.3 Design for Scalability 29217.4 Design for Extensibility 29217.5 Design for Failure 29317.6 Planning Considerations 29417.7 Evolving Traditional Applications 29617.8 Concluding Remarks 301Abbreviations 303References 307About the Authors 311Index 313