设计数据密集型应用（影印版） PDF下载

Part I. Foundations of Data Systems 
1. Reliable, Scalable, and Maintainable Applications 3 
Thinking About Data Systems 4 
Reliability 6 
Hardware Faults 7 
Software Errors 8 
Human Errors 9 
How Important Is Reliability? 10 
Scalability 10 
Describing Load 11 
Describing Performance 13 
Approaches for Coping with Load 17 
Maintainability 18 
Operability: Making Life Easy for Operations 19Part I. Foundations of Data Systems  <；br />；1. Reliable, Scalable, and Maintainable Applications 3  <；br />；Thinking About Data Systems 4  <；br />；Reliability 6  <；br />；Hardware Faults 7  <；br />；Software Errors 8  <；br />；Human Errors 9  <；br />；How Important Is Reliability? 10  <；br />；Scalability 10  <；br />；Describing Load 11  <；br />；Describing Performance 13  <；br />；Approaches for Coping with Load 17  <；br />；Maintainability 18  <；br />；Operability: Making Life Easy for Operations 19  <；br />；Simplicity: Managing Complexity 20  <；br />；Evolvability: Making Change Easy 21  <；br />；Summary 22  <；br />；2. Data Models and Query Languages 27  <；br />；Relational Model Versus Document Model 28  <；br />；The Birth of NoSQL 29  <；br />；The Object-Relational Mismatch 29  <；br />；Many-to-One and Many-to-Many Relationships 33  <；br />；Are Document Databases Repeating History? 36  <；br />；Relational Versus Document Databases Today 38  <；br />；Query Languages for Data 42  <；br />；Declarative Queries on the Web 44  <；br />；MapReduce Querying 46  <；br />；Graph-Like Data Models 49  <；br />；Property Graphs 50  <；br />；The Cypher Query Language 52  <；br />；Graph Queries in SQL 53  <；br />；Triple-Stores and SPARQL 55  <；br />；The Foundation: Datalog 60  <；br />；Summary 63  <；br />；3. Storage and Retrieval 69  <；br />；Data Structures That Power Your Database 70  <；br />；Hash Indexes 72  <；br />；SSTables and LSM-Trees 76  <；br />；B-Trees 79  <；br />；Comparing B-Trees and LSM-Trees 83  <；br />；Other Indexing Structures 85  <；br />；Transaction Processing or Analytics? 90  <；br />；Data Warehousing 91  <；br />；Stars and Snowflakes: Schemas for Analytics 93  <；br />；Column-Oriented Storage 95  <；br />；Column Compression 97  <；br />；Sort Order in Column Storage 99  <；br />；Writing to Column-Oriented Storage 101  <；br />；Aggregation: Data Cubes and Materialized Views 101  <；br />；Summary 103  <；br />4. Encoding and Evolution 111  <br />Formats for Encoding Data 112  <br />Language-Specific Formats 113  <br />JSON, XML, and Binary Variants 114  <br />Thrift and Protocol Buffers 117  <br />Avro 122  <br />The Merits of Schemas 127  <br />Modes of Dataflow 128  <br />Dataflow Through Databases 129  <br />Dataflow Through Services: REST and RPC 131  <br />Message-Passing Dataflow 136  <br />Summary 139  <br />Part II. Distributed Data  <br />5. Replication 151  <br />Leaders and Followers 152  <br />Synchronous Versus Asynchronous Replication 153  <br />Setting Up New Followers 155  <br />Handling Node Outages 156  <br />Implementation of Replication Logs 158  <br />Problems with Replication Lag 161  <br />Reading Your Own Writes 162  <br />Monotonic Reads 164  <br />Consistent Prefix Reads 165  <br />Solutions for Replication Lag 167  <br />Multi-Leader Replication 168  <br />Use Cases for Multi-Leader Replication 168  <br />Handling Write Conflicts 171  <br />Multi-Leader Replication Topologies 175  <br />Leaderless Replication 177  <br />Writing to the Database When a Node Is Down 177  <br />Limitations of Quorum Consistency 181  <br />Sloppy Quorums and Hinted Handoff 183  <br />Detecting Concurrent Writes 184  <br />Summary 192  <br />6. Partitioning 199  <br />Partitioning and Replication 200  <br />Partitioning of Key-Value Data 201  <br />Partitioning by Key Range 202  <br />Partitioning by Hash of Key 203  <br />Skewed Workloads and Relieving Hot Spots 205  <br />Partitioning and Secondary Indexes 206  <br />Partitioning Secondary Indexes by Document 206  <br />Partitioning Secondary Indexes by Term 208  <br />Rebalancing Partitions 209  <br />Strategies for Rebalancing 210  <br />Operations: Automatic or Manual Rebalancing 213  <br />Request Routing 214  <br />Parallel Query Execution 216  <br />Summary 216  <br />7. Transactions 221  <br />The Slippery Concept of a Transaction 222  <br />The Meaning of ACID 223  <br />Single-Object and Multi-Object Operations 228  <br />Weak Isolation Levels 233  <br />Read Committed 234  <br />Snapshot Isolation and Repeatable Read 237  <br />Preventing Lost Updates 242  <br />Write Skew and Phantoms 246  <br />Serializability 251  <br />Actual Serial Execution 252  <br />Two-Phase Locking (2PL) 257  <br />Serializable Snapshot Isolation (SSI) 261  <br />Summary 266  <br />8. The Trouble with Distributed Systems 273  <br />Faults and Partial Failures 274  <br />Cloud Computing and Supercomputing 275  <br />Unreliable Networks 277  <br />Network Faults in Practice 279  <br />Detecting Faults 280  <br />Timeouts and Unbounded Delays 281  <br />Synchronous Versus Asynchronous Networks 284  <br />Unreliable Clocks 287  <br />Monotonic Versus Time-of-Day Clocks 288  <br />Clock Synchronization and Accuracy 289  <br />Relying on Synchronized Clocks 291  <br />Process Pauses 295  <br />Knowledge, Truth, and Lies 300  <br />The Truth Is Defined by the Majority 300  <br />Byzantine Faults 304  <br />System Model and Reality 306  <br />Summary 310  <br />9. Consistency and Consensus 321  <br />Consistency Guarantees 322  <br />Linearizability 324  <br />What Makes a System Linearizable? 325  <br />Relying on Linearizability 330  <br />Implementing Linearizable Systems 332  <br />The Cost of Linearizability 335  <br />Ordering Guarantees 339  <br />Ordering and Causality 339  <br />Sequence Number Ordering 343  <br />Total Order Broadcast 348  <br />Distributed Transactions and Consensus 352  <br />Atomic Commit and Two-Phase Commit (2PC) 354  <br />Distributed Transactions in Practice 360  <br />Fault-Tolerant Consensus 364  <br />Membership and Coordination Services 370  <br />Summary 373  <br />Part III. Derived Data  <br />10. Batch Processing 389  <br />Batch Processing with Unix Tools 391  <br />Simple Log Analysis 391  <br />The Unix Philosophy 394  <br />MapReduce and Distributed Filesystems 397  <br />MapReduce Job Execution 399  <br />Reduce-Side Joins and Grouping 403  <br />Map-Side Joins 408  <br />The Output of Batch Workflows 411  <br />Comparing Hadoop to Distributed Databases 414  <br />Beyond MapReduce 419  <br />Materialization of Intermediate State 419  <br />Graphs and Iterative Processing 424  <br />High-Level APIs and Languages 426  <br />Summary 429  <br />11. Stream Processing 439  <br />Transmitting Event Streams 440  <br />Messaging Systems 441  <br />Partitioned Logs 446  <br />Databases and Streams 451  <br />Keeping Systems in Sync 452  <br />Change Data Capture 454  <br />Event Sourcing 457  <br />State, Streams, and Immutability 459  <br />Processing Streams 464  <br />Uses of Stream Processing 465  <br />Reasoning About Time 468  <br />Stream Joins 472  <br />Fault Tolerance 476  <br />Summary 479  <br />12. The Future of Data Systems 489  <br />Data Integration 490  <br />Combining Specialized Tools by Deriving Data 490  <br />Batch and Stream Processing 494  <br />Unbundling Databases 499  <br />Composing Data Storage Technologies 499  <br />Designing Applications Around Dataflow 504  <br />Observing Derived State 509  <br />Aiming for Correctness 515  <br />The End-to-End Argument for Databases 516  <br />Enforcing Constraints 521  <br />Timeliness and Integrity 524  <br />Trust, but Verify 528  <br />Doing the Right Thing 533  <br />Predictive Analytics 533  <br />Privacy and Tracking 536  <br />Summary 543  <br />Glossary  553  <br />Index 559  <br /><br />