The Definitive Guide to Terracotta: Cluster the JVM for Spring, Hibernate and POJO Scalability

The Definitive Guide to Terracotta: Cluster the JVM for Spring, Hibernate, and POJO Scalability......Page 1
Contents at a Glance......Page 5
Contents......Page 7
Introduction......Page 21
How This Book Is Structured......Page 22
Contacting the Authors......Page 23
Theory and Foundation Forming a Common Understanding......Page 25
Definition of the Terracotta Framework......Page 26
Network Attached Storage Similarities to Terracotta......Page 27
A Layer of Abstraction......Page 28
Parallels to Terracotta......Page 29
Transparency of Memory Location......Page 30
Putting Transparency and Clustering Together......Page 32
Honoring the Java Memory Model......Page 35
Availability......Page 36
Scalability......Page 37
Avoiding Bottlenecks......Page 38
Distributed Caching......Page 41
Session Replication......Page 42
Workload Partitioning......Page 43
Summary......Page 44
History of Terracotta......Page 47
Scale the Database......Page 49
In-Memory Replication......Page 52
All Paths Lead to Partitioning......Page 54
Terracotta’s Origin in a Web Monster......Page 55
The Application Is a Giant Computer......Page 56
Original L2 Was Intrusive......Page 57
Transparency Equals General Purpose......Page 59
Transparent Scalability with Availability......Page 60
Transparent Clustering Service Revisited......Page 61
Clustered Application Servers and Clustered Caches......Page 62
Summary......Page 63
Hello Clustered World......Page 65
Clustering HelloClusteredWorld......Page 67
Why HelloClusteredWorld Works......Page 70
Alternatives to Terracotta......Page 73
A Tour Through Terracotta......Page 74
Terracotta Installation Layout......Page 75
Terracotta Scripts......Page 76
Creating a Project Skeleton......Page 77
Creating HelloClusteredWorld.java......Page 78
Making the Terracotta Server Run in Persistent Mode......Page 79
Configuring Roots......Page 80
Configuring Locks......Page 81
Reviewing the Configuration......Page 82
Starting the Terracotta Server......Page 83
Viewing Runtime Data in the Terracotta Administration Console......Page 84
Heap Durability......Page 86
Step-by-Step Hello Clustered World in Eclipse......Page 87
Implications for Every Developer......Page 89
Summary......Page 90
All About Objects and Roots......Page 93
Virtual Heap and Memory Management......Page 95
Managing Object Changes and Coordinating Threads......Page 96
Locks......Page 97
Autolocks and Named Locks......Page 98
Lock Level......Page 99
Distributed Method Invocation......Page 100
Transparency and Bytecode Instrumentation......Page 101
Physically vs. Logically Managed Objects......Page 102
Nonportable and Logically Managed Classes......Page 103
Transient Fields and On-Load Behavior......Page 104
Initializing Transient Fields on Load......Page 105
Clustered POJOs Step by Step......Page 106
What’s a Clustered POJO?......Page 107
Why Clustered POJOs?......Page 108
The ProductImpl Class......Page 109
The Catalog Class......Page 111
The ShoppingCartImpl Class......Page 112
The Roots Class......Page 113
The Main Class......Page 114
Configuration for the Example......Page 118
Running the Example......Page 119
What Just Happened......Page 123
What the Example Shows......Page 124
Summary......Page 126
The Pain of Large Data Sets......Page 127
The Pain of Duplication......Page 128
How Terracotta Can Help......Page 129
A Simple Example of Caching......Page 131
Transparent Distributed Caching......Page 133
Sorted Maps......Page 136
Autolocked Collections......Page 137
Using Autolocked Collections......Page 138
Bulk Loading a Map......Page 139
Maps of Maps......Page 143
Eviction and Expiration......Page 144
Partitioned Data......Page 145
Leveraging Terracotta......Page 146
Rewriting the Cache Example......Page 147
Enhancements and Configuration......Page 151
Using Ehcache with Terracotta......Page 152
Summary......Page 153
Object-Relational Mapping......Page 155
How Does Hibernate Work?......Page 156
How Terracotta Improves Hibernate Performance......Page 157
Baseline Performance......Page 158
Add Hibernate to Save Three DB Calls......Page 159
Enable the Second-Level Cache to Save Five SQL Calls......Page 160
Clustering the Session Objects......Page 161
Configuring Hibernate to Work with Terracotta......Page 163
Configuring Terracotta......Page 164
Configuring Terracotta to Use the Hibernate Integration Module......Page 165
Storing the Detached Instances......Page 166
Debugging, Testing, and Tuning Concerns......Page 167
Tuning the Second-Level Cache......Page 168
Summary......Page 169
What Are HTTP Sessions?......Page 171
Server-Side State Management......Page 172
Automatic Time-Out of Stale Data......Page 173
Events to Coordinate State Modification......Page 174
Distributed HTTP Sessions......Page 175
Dynamic Horizontal Scale-Out to Handle Traffic Spikes......Page 176
Alternatives to HTTP Sessions......Page 178
How Terracotta Sessions Work......Page 179
Appropriate Lock Granularity and Isolation......Page 180
Rolling Upgrades......Page 181
Simultaneous Linear Scalability and High Availability......Page 182
No Need to Set Attributes After Each Change......Page 183
Shared Session State......Page 184
Package and Configure the Web Application......Page 185
Install and Run the Application with the Terracotta Sessions Configurator......Page 186
Trying the Example......Page 188
HTTP Sessions with Structured Session Attributes......Page 189
Packaging and Configuring the Web Application......Page 193
Manually Configuring and Installing the Example into Tomcat......Page 194
Verifying Fine-Grain Change Detection......Page 195
Understanding HTTP Sessions with Terracotta......Page 196
The Effects of a Round-Robin Load Balancer......Page 197
Using a Sticky Load Balancer......Page 198
Summary......Page 199
Clustering Spring......Page 201
A Sample: Spring Beans Without a Database......Page 202
Running the Example on a Single JVM......Page 211
Running the Example Clustered with Terracotta......Page 214
A Note on DI and Terracotta......Page 217
Growing Spring Beans......Page 218
Clustering Spring Beans......Page 221
Running with Clustered Spring Beans......Page 224
Summary......Page 225
Clustering a Synchronized Map......Page 227
Importing a TIM......Page 228
Module Repositories......Page 229
The Terracotta Forge......Page 230
Creating a Manifest......Page 231
Terracotta Configuration......Page 233
Creating a Simple TIM......Page 234
Creating TIMs with Maven......Page 235
Including Code in a TIM......Page 237
Defining a Bundle Activator......Page 238
Class Replacement and Modification......Page 239
Using Maven to Create TIMs with Code......Page 242
Summary......Page 244
Terracotta and Thread Coordination......Page 245
Thread Coordination Using Concurrency Primitives......Page 246
Thread Coordination Using the java.util.concurrent Package......Page 251
Thread Coordination and JMX Cluster Events......Page 254
Thread Coordination with Queues......Page 259
Summary......Page 265
Compute Grids vs. Data Grids......Page 267
How Grids Handle Failover and High Availability......Page 268
Introducing the Master/Worker Pattern......Page 269
Using the Java Language’s Thread Coordination Primitives......Page 270
Using the java.util.concurrent Abstractions......Page 271
Using the CommonJ Work Manager Specification......Page 272
Implementing a Single-JVM Naïve Master/Worker Container......Page 274
What About My Work?......Page 279
Clustering Our Master/Worker with Terracotta......Page 281
Handling Real-World Challenges......Page 282
Minimizing Contention and Maximizing Locality of Reference......Page 283
Routing Strategies......Page 292
Dealing with Work Failure and Recovery......Page 294
Refactoring the Worker......Page 295
Using the Master/Worker System......Page 296
Work Batching......Page 298
Dynamic Worker Management......Page 299
Worker Failover......Page 303
Building a Distributed Web Spider......Page 305
When Not to Use the Master/Worker Pattern......Page 314
Summary......Page 315
Observing an Application......Page 317
Visualization Tools and Architecture......Page 318
The Sample Inventory Application......Page 319
Modifying to Generate Load......Page 321
Pounder Methods......Page 322
Tuning the Locks......Page 325
Using Separate Roots......Page 327
Updating the CLI......Page 329
Updating the Terracotta Configuration......Page 330
Taking Performance Snapshots......Page 331
Recalling the Motherboard......Page 334
SEDA......Page 338
Terracotta’s SEDA Pipeline......Page 339
Broadcasting Changes......Page 340
Visualizing the Bottlenecks......Page 341
Managing Production Environments......Page 344
Summary......Page 345
Index......Page 347