第0 章 当谈论大数据时我们在谈什么................ 1
0.1 大数据是什么.......................... 2
0.2 大数据之翼:技术范型转换......................................... 4
0.3 大数据商业炼金术................................ 6
0.4 “大数据”在路上................................................... 7
第1 章 数据分片与路由.............................................. 9
1.1 抽象模型.......................................................10
1.2 哈希分片(Hash Partition) ..............................11
1.2.1 Round Robin....................................11
1.2.2 虚拟桶(Virtual Buckets) ..........................12
1.2.3 一致性哈希(Consistent Hashing) ...........................13
1.3 范围分片(Range Partition) ......................................18
参考文献......................................19
第2 章 数据复制与一致性................................................20
2.1 基本原则与设计理念............................21
2.1.1 原教旨CAP 主义..............................................21
2.1.2 CAP 重装上阵(CAP Reloaded).............................23
2.1.3 ACID 原则...............................................24
2.1.4 BASE 原则.................................................24
2.1.5 CAP/ACID/BASE 三者的关系...........................25
2.1.6 幂等性(Idempotent)........................................26
2.2 一致性模型分类.................................................26
2.2.1 强一致性............................................27
2.2.2 最终一致性........................................28
2.2.3 因果一致性.............................28
2.2.4 “读你所写”一致性....................................29
2.2.5 会话一致性....................................29
2.2.6 单调读一致性..............................................30
2.2.7 单调写一致性.....................................................30
2.3 副本更新策略...........................30
2.3.1 同时更新..........................................30
2.3.2 主从式更新.....................................31
2.3.3 任意节点更新......................................32
2.4 一致性协议...........................................................32
2.4.1 两阶段提交协议(Two-Phrase Commit,2PC)..........................33
2.4.2 向量时钟(Vector Clock) ..............................38
2.4.3 RWN 协议.................................................40
2.4.4 Paxos 协议.............................................42
2.4.5 Raft 协议.............................................45
参考文献................................................49
第3 章 大数据常用的算法与数据结构....................................51
3.1 布隆过滤器(Bloom Filter) ............................51
3.1.1 基本原理.............................................52
3.1.2 误判率及相关计算..........................................52
3.1.3 改进:计数Bloom Filter....................................53
3.1.4 应用............................................54
3.2 SkipList............................................55
3.3 LSM 树........................................58
3.4 Merkle 哈希树(Merkle Hash Tree) .............................62
3.4.1 Merkle 树基本原理..................................................62
3.4.2 Dynamo 中的应用.........................................63
3.4.3 比特币中的应用..................................................63
3.5 Snappy 与LZSS 算法..........................................65
3.5.1 LZSS 算法.............................................65
3.5.2 Snappy..........................................67
3.6 Cuckoo 哈希(Cuckoo Hashing) ..................................67
3.6.1 基本原理...............................................68
3.6.2 应用:SILT 存储系统.........................................68
参考文献...................................................70
第4 章 集群资源管理与调度.......................................71
4.1 资源管理抽象模型...................................72
4.1.1 概念模型....................................72
4.1.2 通用架构...............................................73
4.2 调度系统设计的基本问题.....................................74
4.2.1 资源异质性与工作负载异质性............................74
4.2.2 数据局部性(Data Locality) ........................................75
4.2.3 抢占式调度与非抢占式调度...................................75
4.2.4 资源分配粒度(Allocation Granularity) .............76
4.2.5 饿死(Starvation)与死锁(Dead Lock)问题...........................76
4.2.6 资源隔离方法........................................77
4.3 资源管理与调度系统范型.............................77
4.3.1 集中式调度器(Monolithic Scheduler).......................78
4.3.2 两级调度器(Two-Level Scheduler) .........................79
4.3.3 状态共享调度器(Shared-State Scheduler) ....................79
4.4 资源调度策略...............................................81
4.4.1 FIFO 调度策略..........................................81
4.4.2 公平调度器(Fair Scheduler)......................81
4.4.3 能力调度器(Capacity Scheduler) ..........................82
4.4.4 延迟调度策略(Delay Scheduling)............................82
4.4.5 主资源公平调度策略(Dominant Resource Fair Scheduling).............82
4.5 Mesos .................................84
4.6 YARN......................................87
参考文献..............................................90
第5 章 分布式协调系统...................................91
5.1 Chubby 锁服务...............................92
5.1.1 系统架构........................................93
5.1.2 数据模型..................................94
5.1.3 会话与KeepAlive 机制...............................95
5.1.4 客户端缓存.......................................95
5.2 ZooKeeper ................................96
5.2.1 体系结构...........................................96
5.2.2 数据模型(Data Model) .............................97
5.2.3 API ...............................98
5.2.4 ZooKeeper 的典型应用场景..................................98
5.2.5 ZooKeeper 的实际应用.......................................103
参考文献...................................104
第6 章 分布式通信..............................106
6.1 序列化与远程过程调用框架..................................107
6.1.1 Protocol Buffer 与Thrift .....................108
6.1.2 Avro...............................109
6.2 消息队列.....................................110
6.2.1 常见的消息队列系统......................................110
6.2.2 Kafka .......................111
6.3 应用层多播通信(Application-Level Multi-Broadcast)........114
6.3.1 概述...............................114
6.3.2 Gossip 协议...........................115
参考文献..........................118
第7 章 数据通道.........................................120
7.1 Log 数据收集.................................120
7.1.1 Chukwa........................121
7.1.2 Scribe......................122
7.2 数据总线......................................123
7.2.1 Databus............................125
7.2.2 Wormhole .......................127
7.3 数据导入/导出...........................................128
参考文献.............................129
第8 章 分布式文件系统....................................131
8.1 Google 文件系统(GFS) .................................132
8.1.1 GFS 设计原则...........................................132
8.1.2 GFS 整体架构..............................133
8.1.3 GFS 主控服务器..................................134
8.1.4 系统交互行为.................................136
8.1.5 Colossus ........................137
8.2 HDFS ..........................138
8.2.1 HDFS 整体架构.................................139
8.2.2 HA 方案..............................140
8.2.3 NameNode 联盟........................143
8.3 HayStack 存储系统....................................145
8.3.1 HayStack 整体架构.................................146
8.3.2 目录服务..................................147
8.3.3 HayStack 缓存...........................................148
8.3.4 HayStack 存储系统的实现...............................148
8.4 文件存储布局.........................................150
8.4.1 行式存储........................................151
8.4.2 列式存储...........................................151
8.4.3 混合式存储........................................156
8.5 纠删码(Erasure Code).............................158
8.5.1 Reed-Solomon 编码...............................159
8.5.2 LRC 编码.....................................164
8.5.3 HDFS-RAID 架构.........................166
参考文献.....................................166
第9 章 内存KV 数据库...................................168
9.1 RAMCloud ..............................169
9.1.1 RAMCloud 整体架构................................169
9.1.2 数据副本管理与数据恢复................................170
9.2 Redis....................................172
9.3 MemBase ...............................173
参考文献................................................175
第10 章 列式数据库...........................................176
10.1 BigTable....................................177
10.1.1 BigTable 的数据模型..........................177
10.1.2 BigTable 的整体结构................................178
10.1.3 BigTable 的管理数据.............................179
10.1.4 主控服务器(Master Server)......................181
10.1.5 子表服务器(Tablet Server) ....................182
10.2 PNUTS 存储系统........................................186
10.2.1 PNUTS 的整体架构..............................186
10.2.2 存储单元...............................187
10.2.3 子表控制器与数据路由器..................................187
10.2.4 雅虎消息代理.............................188
10.2.5 数据一致性.........................................189
10.3 MegaStore..................................................190
10.3.1 实体群组切分......................191
10.3.2 数据模型........................................192
10.3.3 数据读/写与备份.................................193
10.4 Spanner .........................................194
10.4.1 SpanServer 软件栈.........................................195
10.4.2 数据模型.........................................196
10.4.3 TrueTime ...........................................196
参考文献..............................................197
第11 章 大规模批处理系统...................................199
11.1 MapReduce 计算模型与架构................................200
11.1.1 计算模型.......................................201
11.1.2 系统架构......................................203
11.1.3 MapReduce 计算的特点及不足......................................206
11.2 MapReduce 计算模式...........................206
11.2.1 求和模式(Summarization Pattern)................207
11.2.2 过滤模式(Filtering Pattern) ................208
11.2.3 组织数据模式(Data Organization Pattern) .....................210
11.2.4 Join 模式(Join Pattern)......................212
11.3 DAG 计算模型..........................................214
11.3.1 DAG 计算系统的三层结构............................214
11.3.2 Dryad .......................................215
11.3.3 FlumeJava 和Tez ........................................217
参考文献...........................................218
第12 章 流式计算........................................219
12.1 流式计算系统架构....................................222
12.1.1 主从架构............................................222
12.1.2 P2P 架构.....................................................223
12.1.3 Samza 架构..........................................224
12.2 DAG 拓扑结构..........................................224
12.2.1 计算节点.....................................................225
12.2.2 数据流..............................................226
12.2.3 拓扑结构..................................226
12.3 送达保证(Delivery Guarantees)..............................229
12.3.1 Storm 的送达保证机制.................................230
12.3.2 MillWheel 的“恰好送达一次”机制...........................233
12.4 状态持久化...........................................234
12.4.1 容错的三种模式....................................234
12.4.2 Storm 的状态持久化.......................................236
12.4.3 MillWheel 和Samza 的状态持久化......................237
参考文献............................................238
第13 章 交互式数据分析...................................240
13.1 Hive 系数据仓库.................................242
13.1.1 Hive .....................................242
13.1.2 StingerInitiative ................................250
13.2 Shark 系数据仓库..................................251
13.2.1 Shark 架构.........................................252
13.2.2 部分DAG 执行引擎(PDE) ........................253
13.2.3 数据共同分片.........................................254
13.3 Dremel 系数据仓库...................................254
13.3.1 Dremel...........................255
13.3.2 PowerDrill ..........................258
13.3.3 Impala.................................261
13.3.4 Presto...............................264
13.4 混合系数据仓库......................................265
参考文献.................................269
第14 章 图数据库:架构与算法................................271
14.1 在线查询类图数据库...........................272
14.1.1 三层结构.........................272
14.1.2 TAO 图数据库.................................273
14.2 常见图挖掘问题..........................................277
14.2.1 PageRank 计算.......................................278
14.2.2 单源最短路径(Single Source Shortest Path) ..................278
14.2.3 二部图最大匹配.............................279
14.3 离线挖掘数据分片..............................................279
14.3.1 切边法(Edge-Cut)......................................280
14.3.2 切点法(Vertex-Cut)...............................282
14.4 离线挖掘计算模型...................................284
14.4.1 以节点为中心的编程模型..........................284
14.4.2 GAS 编程模型...........................................285
14.4.3 同步执行模型.....................................286
14.4.4 异步执行模型...................................290
14.5 离线挖掘图数据库.................................292
14.5.1 Pregel..........................292
14.5.2 Giraph...............................299
14.5.3 GraphChi ............................301
14.5.4 PowerGraph.......................307
参考文献.......................................311
第15 章 机器学习:范型与架构.........................................313
15.1 分布式机器学习...........................................314
15.1.1 机器学习简介.............................................314
15.1.2 数据并行VS.模型并行.....................................316
15.2 分布式机器学习范型.....................317
15.2.1 三种范型...................................318
15.2.2 MapReduce 迭代计算模型........................319
15.2.3 BSP 计算模型...................................321
15.2.4 SSP 模型............................323
15.3 分布式机器学习架构...................................324
15.3.1 MapReduce 系列..................................325
15.3.2 Spark 及MLBase ..........................................327
15.3.3 参数服务器(Parameter Server).............332
参考文献................................................335
第16 章 机器学习:分布式算法...............................337
16.1 计算广告:逻辑回归.......................................338
16.1.1 逻辑回归(Logistic Regression,LR).............................338
16.1.2 并行随机梯度下降(Parallel Stochastic Gradient Descent)............341
16.1.3 批学习并行逻辑回归..................................341
16.2 推荐系统:矩阵分解................................................344
16.2.1 矩阵分解方法.......................................344
16.2.2 ALS-WR 算法............................................346
16.2.3 并行版ALS-WR 算法..............................347
16.3 搜索引擎:机器学习排序................................347
16.3.1 机器学习排序简介.................................348
16.3.2 LambdaMART.................................349
16.3.3 分布式LambdaMART........................................351
16.4 自然语言处理:文档相似性计算.......................................352
16.5 社交挖掘:谱聚类.................................355
16.5.1 社交挖掘实例...............................355
16.5.2 谱聚类....................................356
16.5.3 并行版谱聚类..........................................358
16.6 深度学习:DistBelief .............................................358
16.6.1 深度学习简介........................................359
16.6.2 DistBelief.....................360
参考文献.........................................364
第17 章 增量计算..........................................366
17.1 增量计算模式...........................367
17.1.1 两种计算模式...............................367
17.1.2 Hadoop 平台下增量计算的一般模式.............................368
17.2 Percolator................................370
17.2.1 事务支持..........................................371
17.2.2 “观察/通知”体系结构...........................373
17.3 Kineograph ............................374
17.3.1 整体架构.........................................375
17.3.2 增量计算机制....................................375
17.4 DryadInc ....................................376
参考文献..............................................................377
附录A 硬件体系结构及常用性能指标......................................378
附录B 大数据必读文献....................................380
· · · · · · (
收起)