| |
| |
| Introduction | |
| |
| |
| |
| Scaling Up (and Down) | |
| |
| |
| Defining Scalability | |
| |
| |
| Why Do We Need to Scale? | |
| |
| |
| Scaling Up Gone Wrong | |
| |
| |
| What Goes Up Should Go Down | |
| |
| |
| Real-World Considerations | |
| |
| |
| Being a Good Architect | |
| |
| |
| |
| Principles for Avoiding Failure | |
| |
| |
| Working in Production Environments | |
| |
| |
| The Dot Bomb and Its Effect on Building Large Systems | |
| |
| |
| Stability and Control | |
| |
| |
| Rapid Development | |
| |
| |
| Unit Testing | |
| |
| |
| Version Control | |
| |
| |
| Version Control in Action | |
| |
| |
| A Different Approach to Disaster Recovery | |
| |
| |
| Good Design | |
| |
| |
| |
| Mission-Critical Environments | |
| |
| |
| High Availability | |
| |
| |
| Monitoring | |
| |
| |
| Monitoring Implementations | |
| |
| |
| Criteria for a Capable Monitoring System | |
| |
| |
| Coping with Release Cycles | |
| |
| |
| Internal Release Cycles | |
| |
| |
| External Release Cycles | |
| |
| |
| The Cost of Complexity Shackled to Large Architectures | |
| |
| |
| Looking for Speed | |
| |
| |
| It's Not a One-Man Mission | |
| |
| |
| |
| High Availability. HA! No Downtime?! | |
| |
| |
| Why High Availability and Load Balancing Are Different | |
| |
| |
| Load Balancing Is Not High Availability | |
| |
| |
| High Availability Is Not Load Balancing | |
| |
| |
| Traditional High Availability-The Whitepaper Approach | |
| |
| |
| Surveying the Site | |
| |
| |
| Pouring Concrete-Foundry ServerIron | |
| |
| |
| High Availability Rethought (Peer-Based) | |
| |
| |
| Peer-Based High Availability in the Real World | |
| |
| |
| Growing Seamlessly | |
| |
| |
| Diverging from the Web: High Availability for Email, DNS, and So On | |
| |
| |
| Pouring Concrete | |
| |
| |
| |
| Load Balancing and the Utter Confusion Surrounding It | |
| |
| |
| Load Balancing Basics | |
| |
| |
| IP-Friendly Services | |
| |
| |
| Web Switches | |
| |
| |
| IP Virtual Servers | |
| |
| |
| Application Layer Load Balancers | |
| |
| |
| A Real Use | |
| |
| |
| Abusing the Term | |
| |
| |
| The Web and Beyond | |
| |
| |
| Session Stickiness: A Fatal Misconception | |
| |
| |
| So, Which One Is the Best? | |
| |
| |
| |
| Static Content Serving for Speed and Glory | |
| |
| |
| What's So Tricky? | |
| |
| |
| Context Switching | |
| |
| |
| Resource Utilization | |
| |
| |
| The Goal | |
| |
| |
| Building the Cluster | |
| |
| |
| Setting Goals | |
| |
| |
| Tackling Content Distribution | |
| |
| |
| A Priori Placement | |
| |
| |
| Cache-on-Demand | |
| |
| |
| Choosing a Web Serving Platform | |
| |
| |
| Examining Our Availability Requirements | |
| |
| |
| Choosing an OS | |
| |
| |
| Wackamole | |
| |
| |
| Reasoning | |
| |
| |
| Installation | |
| |
| |
| Testing the High Availability | |
| |
| |
| Serving Content | |
| |
| |
| World Domination | |
| |
| |
| DNS Round-Trip Times | |
| |
| |
| Anycast-Shared IP | |
| |
| |
| A Final Review | |
| |
| |
| Was It Worth It? | |
| |
| |
| |
| Static Meets Dynamic Adding Caches to Reduce Costs | |
| |
| |
| Types of Caches | |
| |
| |
| Layered/Transparent Cache | |
| |
| |
| Integrated (Look-Aside) Cache | |
| |
| |
| Data Cache | |
| |
| |
| Write-Thru and Write-Back Caches | |
| |
| |
| Distributed Cache | |
| |
| |
| Deploying Caches | |
| |
| |
| Serving a News Site | |
| |
| |
| Simple Implementation | |
| |
| |
| Introducing Integrated Caching | |
| |
| |
| Tackling User Data | |
| |
| |
| Two-Tier Execution | |
| |
| |
| Caching Is Powerful | |
| |
| |
| |
| Distributed Databases Are Easy, Just Read the Fine Print | |
| |
| |
| What Is a Distributed Database? | |
| |
| |
| Data Resiliency | |
| |
| |
| Operational Failover | |
| |
| |
| Increased Query Performance | |
| |
| |
| Complete Reliability | |
| |
| |
| Geographically Distributed Operation | |
| |
| |
| Why Is Replication So Hard? | |
| |
| |
| Multimaster Replication | |
| |
| |
| Two-Phase Commit (2PC) | |
| |
| |
| EVS Engine | |
| |
| |
| Master-Master Replication | |
| |
| |
| Master-Slave Replication | |
| |
| |
| Changing the Scope of the Problem | |
| |
| |
| Operation and Changeset Replication | |
| |
| |
| News Site Revisited | |
| |
| |
| Choosing Technologies and Methods | |
| |
| |
| Implementing Cross-Vendor Database Replication | |
| |
| |
| Same Vendor Database Replication Is Easy | |
| |
| |
| |
| Juggling Logs and Other Circus Tricks | |
| |
| |
| Why Is Logging a Challenge? | |
| |
| |
| Classic Solutions | |
| |
| |
| Periodic "Batch" Aggregation | |
| |
| |
| Real-time Unicast Aggregation | |
| |
| |
| Passive "Sniffing" Log Aggregation | |
| |
| |
| Logging Done Right | |
| |
| |
| The Architecture (Servers, Software, and Roles) | |
| |
| |
| Building It | |
| |
| |
| Subscribers-spreadlogd | |
| |
| |
| Publishers-mod_log_spread | |
| |
| |
| Understanding the Beauty of the Beast | |
| |
| |
| Real-time Analysis | |
| |
| |
| Real-time Monitoring | |
| |
| |
| Passive Log Aggregation for Metrics | |
| |
| |
| Visualizing Data with RRDtool | |
| |
| |
| A Bit About RRDtool | |
| |
| |
| Setting Up Our Databases | |
| |
| |
| Collecting Metrics | |
| |
| |
| Visualizing Data Through RRDtool | |
| |
| |
| Being Hit in the Face with Data | |
| |
| |
| Generic Uses | |
| |
| |
| |
| The Right Tool for the Job | |
| |
| |
| Who's Online? | |
| |
| |
| Technical Setup | |
| |
| |
| Defining Scope | |
| |
| |
| Stepping Back | |
| |
| |
| Thinking Outside the Box | |
| |
| |
| Testing the Solution | |
| |
| |
| Who's Online Reviewed | |
| |
| |
| |
| Spread | |
| |
| |
| Group Communication | |
| |
| |
| Introducing...Spread | |
| |
| |
| Installation and Configuration | |
| |
| |
| Getting Started | |
| |
| |
| A Practical Example | |
| |
| |
| Index | |