SlideShare a Scribd company logo
1 of 71
Intorduction to Distributed Infrastructure




                    Intorduction to Distributed Infrastructure

                                              Guang Feng

                                             dudu@hunantv.com


                                             April 29, 2010
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
                        分布式计算框架
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
                        分布式计算框架
                        分布式存储引擎
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
                        分布式计算框架
                        分布式存储引擎
                数据库会议的一些个人思考
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
                        分布式计算框架
                        分布式存储引擎
                数据库会议的一些个人思考
                        Language Tower
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
                        分布式计算框架
                        分布式存储引擎
                数据库会议的一些个人思考
                        Language Tower
                        数据库性能监控与分析
Intorduction to Distributed Infrastructure




      Topic
                当今分布式技术主要有以下几个研究方向
                        分布式文件系统
                        分布式计算框架
                        分布式存储引擎
                数据库会议的一些个人思考
                        Language Tower
                        数据库性能监控与分析
                        DBA团队建设与职责
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
                        大容量文件存储
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
                        大容量文件存储
                        本地文件系统的I/O瓶颈
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
                        大容量文件存储
                        本地文件系统的I/O瓶颈
                        灾难备份
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
                        大容量文件存储
                        本地文件系统的I/O瓶颈
                        灾难备份
                主要产品
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
                        大容量文件存储
                        本地文件系统的I/O瓶颈
                        灾难备份
                主要产品
                        面向大文件存储 Google GFS like...
Intorduction to Distributed Infrastructure




      Distributed File System
                解决的问题
                        大容量文件存储
                        本地文件系统的I/O瓶颈
                        灾难备份
                主要产品
                        面向大文件存储 Google GFS like...
                        面向小文件存储 MogileFS
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                概念
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                概念
                        MapReduce
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                概念
                        MapReduce
                        并行处理的是数据,不是任务(计算能力)
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                概念
                        MapReduce
                        并行处理的是数据,不是任务(计算能力)
                产品
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                概念
                        MapReduce
                        并行处理的是数据,不是任务(计算能力)
                产品
                        Hadoop
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                概念
                        MapReduce
                        并行处理的是数据,不是任务(计算能力)
                产品
                        Hadoop
                        *Gearman
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                Gearman




                                             Figure: Gearman Arch
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                Gearman
                        work in parallel




                                             Figure: Gearman Arch
Intorduction to Distributed Infrastructure




      Distributed Computing Framework
                Gearman
                        work in parallel
                        background task




                                             Figure: Gearman Arch
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
                        Amazon Dynamo
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
                        Amazon Dynamo
                Not Only SQL
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
                        Amazon Dynamo
                Not Only SQL
                        Tokyo Cabinet
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
                        Amazon Dynamo
                Not Only SQL
                        Tokyo Cabinet
                Relational Database
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
                        Amazon Dynamo
                Not Only SQL
                        Tokyo Cabinet
                Relational Database
                        on-Proxy
Intorduction to Distributed Infrastructure




      Distributed Storage Engine
                Colunmn-Based Storage
                        Google BigTable
                        Amazon Dynamo
                Not Only SQL
                        Tokyo Cabinet
                Relational Database
                        on-Proxy
                        Cluster
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
                        每单元(Tracker)中记录统一数据的多个版本
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
                        每单元(Tracker)中记录统一数据的多个版本
                        数据聚集性强
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
                        每单元(Tracker)中记录统一数据的多个版本
                        数据聚集性强
                缺点
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
                        每单元(Tracker)中记录统一数据的多个版本
                        数据聚集性强
                缺点
                        不适合频繁CRUD
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
                        每单元(Tracker)中记录统一数据的多个版本
                        数据聚集性强
                缺点
                        不适合频繁CRUD
                        不支持条件查询
Intorduction to Distributed Infrastructure




      Colunmn-Based Storage
                优势
                        动态增加存储列,水平切分伸缩性强
                        每单元(Tracker)中记录统一数据的多个版本
                        数据聚集性强
                缺点
                        不适合频繁CRUD
                        不支持条件查询
                        不适合OLTP(On-Line Analytical Processing)
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
                        键值对存储
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
                        键值对存储
                        insert 0.4s/1M records & search 0.33s/1M records
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
                        键值对存储
                        insert 0.4s/1M records & search 0.33s/1M records
                        支持丰富的语言调用(HTTP协议)
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
                        键值对存储
                        insert 0.4s/1M records & search 0.33s/1M records
                        支持丰富的语言调用(HTTP协议)
                不足
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
                        键值对存储
                        insert 0.4s/1M records & search 0.33s/1M records
                        支持丰富的语言调用(HTTP协议)
                不足
                        上亿的数据存储后,性能下降很快
Intorduction to Distributed Infrastructure




      NoSQL:Tokyo Cabinet
                特性
                        键值对存储
                        insert 0.4s/1M records & search 0.33s/1M records
                        支持丰富的语言调用(HTTP协议)
                不足
                        上亿的数据存储后,性能下降很快
                        很难扩展到多机
Intorduction to Distributed Infrastructure




      RDB:Amoeba

                Proxy Architecture(代理风格)
Intorduction to Distributed Infrastructure




      RDB:Amoeba

                Proxy Architecture(代理风格)
                特性
Intorduction to Distributed Infrastructure




      RDB:Amoeba

                Proxy Architecture(代理风格)
                特性
                        提供规则数据切分
Intorduction to Distributed Infrastructure




      RDB:Amoeba

                Proxy Architecture(代理风格)
                特性
                        提供规则数据切分
                        降低db 与客户端的连接数
Intorduction to Distributed Infrastructure




      RDB:Amoeba

                Proxy Architecture(代理风格)
                特性
                        提供规则数据切分
                        降低db 与客户端的连接数
                        读写分离
Intorduction to Distributed Infrastructure




      RDB:Amoeba

                Proxy Architecture(代理风格)
                特性
                        提供规则数据切分
                        降低db 与客户端的连接数
                        读写分离
                        未来打算支持异构数据库
Intorduction to Distributed Infrastructure




      RDB:EnterpriseDB Cluster
                特性
Intorduction to Distributed Infrastructure




      RDB:EnterpriseDB Cluster
                特性
                        内置连接池,内存共享
Intorduction to Distributed Infrastructure




      RDB:EnterpriseDB Cluster
                特性
                        内置连接池,内存共享
                        读请求负载均衡
Intorduction to Distributed Infrastructure




      RDB:EnterpriseDB Cluster
                特性
                        内置连接池,内存共享
                        读请求负载均衡
                        所有节点并行查询
Intorduction to Distributed Infrastructure




      RDB:EnterpriseDB Cluster
                特性
                        内置连接池,内存共享
                        读请求负载均衡
                        所有节点并行查询
                        查询缓存
Intorduction to Distributed Infrastructure




      Language Tower

                分布式基础架构
Intorduction to Distributed Infrastructure




      Language Tower

                分布式基础架构
                语言无关的通讯协议
Intorduction to Distributed Infrastructure




      Language Tower

                分布式基础架构
                语言无关的通讯协议
                动态脚本语言实现业务逻辑
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
                        数据库日志(二进制日志、慢查询)分析
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
                        数据库日志(二进制日志、慢查询)分析
                        根据业务特点进行针对优化
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
                        数据库日志(二进制日志、慢查询)分析
                        根据业务特点进行针对优化
                DBA团队建设与职责
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
                        数据库日志(二进制日志、慢查询)分析
                        根据业务特点进行针对优化
                DBA团队建设与职责
                        日常维护数据
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
                        数据库日志(二进制日志、慢查询)分析
                        根据业务特点进行针对优化
                DBA团队建设与职责
                        日常维护数据
                        Top Query的分析与解决
Intorduction to Distributed Infrastructure




      Thinking...
                数据库性能监控与分析
                        结合主机负载与数据库指标进行监控
                        数据库日志(二进制日志、慢查询)分析
                        根据业务特点进行针对优化
                DBA团队建设与职责
                        日常维护数据
                        Top Query的分析与解决
                        SQL由谁来写?
Intorduction to Distributed Infrastructure




      Thank you!

       Q&A

More Related Content

What's hot

Hadoop development in China Mobile Research Institute
Hadoop development in China Mobile Research InstituteHadoop development in China Mobile Research Institute
Hadoop development in China Mobile Research InstituteXu Wang
 
20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索
20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索
20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索liu sheng
 
周敏奇:Cliaims—集群感知的内存计算系统
周敏奇:Cliaims—集群感知的内存计算系统周敏奇:Cliaims—集群感知的内存计算系统
周敏奇:Cliaims—集群感知的内存计算系统hdhappy001
 
高效益、設計專利保護 如何達成雙贏?
高效益、設計專利保護 如何達成雙贏?高效益、設計專利保護 如何達成雙贏?
高效益、設計專利保護 如何達成雙贏?NVIDIA Taiwan
 
Challenges and opportunities computing Kuo-Yi Chen
Challenges and opportunities computing   Kuo-Yi ChenChallenges and opportunities computing   Kuo-Yi Chen
Challenges and opportunities computing Kuo-Yi Chenkuoyichen
 

What's hot (8)

Hadoop development in China Mobile Research Institute
Hadoop development in China Mobile Research InstituteHadoop development in China Mobile Research Institute
Hadoop development in China Mobile Research Institute
 
20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索
20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索
20141128(刘胜)UTC2014分布式和云服务的思考与实践——支付清算行业分布式架构的探索
 
Symantec System Recovery 2011_0320
Symantec System Recovery 2011_0320Symantec System Recovery 2011_0320
Symantec System Recovery 2011_0320
 
Symantec system recovery 2011產品技術訓練
Symantec system recovery 2011產品技術訓練Symantec system recovery 2011產品技術訓練
Symantec system recovery 2011產品技術訓練
 
Altibase介绍
Altibase介绍Altibase介绍
Altibase介绍
 
周敏奇:Cliaims—集群感知的内存计算系统
周敏奇:Cliaims—集群感知的内存计算系统周敏奇:Cliaims—集群感知的内存计算系统
周敏奇:Cliaims—集群感知的内存计算系统
 
高效益、設計專利保護 如何達成雙贏?
高效益、設計專利保護 如何達成雙贏?高效益、設計專利保護 如何達成雙贏?
高效益、設計專利保護 如何達成雙贏?
 
Challenges and opportunities computing Kuo-Yi Chen
Challenges and opportunities computing   Kuo-Yi ChenChallenges and opportunities computing   Kuo-Yi Chen
Challenges and opportunities computing Kuo-Yi Chen
 

Viewers also liked

Current Construction Projects, City of Chicago, August 21, 2008
Current Construction Projects, City of Chicago, August 21, 2008Current Construction Projects, City of Chicago, August 21, 2008
Current Construction Projects, City of Chicago, August 21, 2008Daniel X. O'Neil
 
Construction Challenge
Construction ChallengeConstruction Challenge
Construction Challengereuther
 
Feasibility of hcl in tr1 copy - copy
Feasibility of hcl in tr1   copy - copyFeasibility of hcl in tr1   copy - copy
Feasibility of hcl in tr1 copy - copyAbhijit Dhaliwal
 
EMC_Final presentation_Neha
EMC_Final presentation_NehaEMC_Final presentation_Neha
EMC_Final presentation_NehaNeha Parekh
 
Top 10 infrastructure project manager interview questions and answers
Top 10 infrastructure project manager interview questions and answersTop 10 infrastructure project manager interview questions and answers
Top 10 infrastructure project manager interview questions and answersfusojom
 
Hadoop on raspberry pi PART 1
Hadoop on raspberry pi PART 1Hadoop on raspberry pi PART 1
Hadoop on raspberry pi PART 1Aniket Maithani
 
Basics of infrastructure project construction raihan uddin ahmed
Basics of infrastructure project construction raihan uddin ahmedBasics of infrastructure project construction raihan uddin ahmed
Basics of infrastructure project construction raihan uddin ahmedRaihan Ahmed
 
Intro infrastructure project 2015.02.26
Intro infrastructure project 2015.02.26Intro infrastructure project 2015.02.26
Intro infrastructure project 2015.02.26Uyanga Nyamkhuu
 
Final year project on Remote Infrastructure Management
Final year project on Remote Infrastructure ManagementFinal year project on Remote Infrastructure Management
Final year project on Remote Infrastructure Managementjairaman
 
Bsnl training by keshav
Bsnl training  by keshavBsnl training  by keshav
Bsnl training by keshavKeshav Katiyar
 
Summer internship project
Summer internship projectSummer internship project
Summer internship projectSunil Yadav
 
Summer internship project report new
Summer internship project report newSummer internship project report new
Summer internship project report newVaibhav Sahu
 
Infrastructure in india ,Indian Infrastructure
Infrastructure in india ,Indian InfrastructureInfrastructure in india ,Indian Infrastructure
Infrastructure in india ,Indian InfrastructureShubham Jain
 
Infrastructure development in india
Infrastructure development in indiaInfrastructure development in india
Infrastructure development in indiaMukesh Khinchi
 
Full summer training report
Full summer training reportFull summer training report
Full summer training reportwadhwa.ritika8
 
Summer internship project hr mba mms
Summer internship project hr mba mmsSummer internship project hr mba mms
Summer internship project hr mba mmsSria
 

Viewers also liked (20)

Current Construction Projects, City of Chicago, August 21, 2008
Current Construction Projects, City of Chicago, August 21, 2008Current Construction Projects, City of Chicago, August 21, 2008
Current Construction Projects, City of Chicago, August 21, 2008
 
Construction Challenge
Construction ChallengeConstruction Challenge
Construction Challenge
 
Week 25 presentation
Week 25 presentationWeek 25 presentation
Week 25 presentation
 
Feasibility of hcl in tr1 copy - copy
Feasibility of hcl in tr1   copy - copyFeasibility of hcl in tr1   copy - copy
Feasibility of hcl in tr1 copy - copy
 
EMC_Final presentation_Neha
EMC_Final presentation_NehaEMC_Final presentation_Neha
EMC_Final presentation_Neha
 
Top 10 infrastructure project manager interview questions and answers
Top 10 infrastructure project manager interview questions and answersTop 10 infrastructure project manager interview questions and answers
Top 10 infrastructure project manager interview questions and answers
 
Hadoop on raspberry pi PART 1
Hadoop on raspberry pi PART 1Hadoop on raspberry pi PART 1
Hadoop on raspberry pi PART 1
 
Basics of infrastructure project construction raihan uddin ahmed
Basics of infrastructure project construction raihan uddin ahmedBasics of infrastructure project construction raihan uddin ahmed
Basics of infrastructure project construction raihan uddin ahmed
 
MBA PROJECT
MBA PROJECTMBA PROJECT
MBA PROJECT
 
Intro infrastructure project 2015.02.26
Intro infrastructure project 2015.02.26Intro infrastructure project 2015.02.26
Intro infrastructure project 2015.02.26
 
Final year project on Remote Infrastructure Management
Final year project on Remote Infrastructure ManagementFinal year project on Remote Infrastructure Management
Final year project on Remote Infrastructure Management
 
Bsnl training by keshav
Bsnl training  by keshavBsnl training  by keshav
Bsnl training by keshav
 
Summer internship project
Summer internship projectSummer internship project
Summer internship project
 
Summer internship project report new
Summer internship project report newSummer internship project report new
Summer internship project report new
 
Infrastructure impact on Indian economy
Infrastructure impact on Indian economyInfrastructure impact on Indian economy
Infrastructure impact on Indian economy
 
Infrastructure in india ,Indian Infrastructure
Infrastructure in india ,Indian InfrastructureInfrastructure in india ,Indian Infrastructure
Infrastructure in india ,Indian Infrastructure
 
Infrastructure development in india
Infrastructure development in indiaInfrastructure development in india
Infrastructure development in india
 
Full summer training report
Full summer training reportFull summer training report
Full summer training report
 
Summer internship project hr mba mms
Summer internship project hr mba mmsSummer internship project hr mba mms
Summer internship project hr mba mms
 
Internship report
Internship reportInternship report
Internship report
 

Similar to 2010 ChinaDatabaseCon "Intorduction to Distributed Infrastructure"

企业系统商务智能设计
企业系统商务智能设计企业系统商务智能设计
企业系统商务智能设计George Ang
 
淘宝网架构变迁和挑战(Oracle架构师日)
淘宝网架构变迁和挑战(Oracle架构师日)淘宝网架构变迁和挑战(Oracle架构师日)
淘宝网架构变迁和挑战(Oracle架构师日)vanadies10
 
Couchbase introduction - Chinese
Couchbase introduction - Chinese Couchbase introduction - Chinese
Couchbase introduction - Chinese Vickie Zeng
 
Dreaming Infrastructure
Dreaming InfrastructureDreaming Infrastructure
Dreaming Infrastructurekyhpudding
 
Big Data, NoSQL, and MongoDB
Big Data, NoSQL, and MongoDBBig Data, NoSQL, and MongoDB
Big Data, NoSQL, and MongoDBMonster Supreme
 
大型视频网站单点分析与可用性提升-Qcon2011
大型视频网站单点分析与可用性提升-Qcon2011大型视频网站单点分析与可用性提升-Qcon2011
大型视频网站单点分析与可用性提升-Qcon2011Yiwei Ma
 
Java 与 云计算
Java 与 云计算Java 与 云计算
Java 与 云计算kevin huang
 
华为软件定义存储架构分析
华为软件定义存储架构分析华为软件定义存储架构分析
华为软件定义存储架构分析Liang Ming
 
众行业公司系统架构案例介绍
众行业公司系统架构案例介绍众行业公司系统架构案例介绍
众行业公司系统架构案例介绍mysqlops
 
设计可扩展的Oracle应用
设计可扩展的Oracle应用设计可扩展的Oracle应用
设计可扩展的Oracle应用james tong
 
深入研究雲端應用程式平台-AppFabric
深入研究雲端應用程式平台-AppFabric深入研究雲端應用程式平台-AppFabric
深入研究雲端應用程式平台-AppFabricJohn Chang
 
基于 MySQL 的B2C电商系统前端数据层架构
基于 MySQL 的B2C电商系统前端数据层架构基于 MySQL 的B2C电商系统前端数据层架构
基于 MySQL 的B2C电商系统前端数据层架构Sky Jian
 
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境drewz lin
 
Etu DW Offload 解放資料倉儲的運算效能
Etu DW Offload 解放資料倉儲的運算效能Etu DW Offload 解放資料倉儲的運算效能
Etu DW Offload 解放資料倉儲的運算效能Etu Solution
 
Zh tw introduction_to_hadoop and hdfs
Zh tw introduction_to_hadoop and hdfsZh tw introduction_to_hadoop and hdfs
Zh tw introduction_to_hadoop and hdfsTrendProgContest13
 
数据访问层开发实践
数据访问层开发实践数据访问层开发实践
数据访问层开发实践xcq
 
Velocity2011分享
Velocity2011分享Velocity2011分享
Velocity2011分享Zoom Quiet
 

Similar to 2010 ChinaDatabaseCon "Intorduction to Distributed Infrastructure" (20)

企业系统商务智能设计
企业系统商务智能设计企业系统商务智能设计
企业系统商务智能设计
 
淘宝网架构变迁和挑战(Oracle架构师日)
淘宝网架构变迁和挑战(Oracle架构师日)淘宝网架构变迁和挑战(Oracle架构师日)
淘宝网架构变迁和挑战(Oracle架构师日)
 
Couchbase introduction - Chinese
Couchbase introduction - Chinese Couchbase introduction - Chinese
Couchbase introduction - Chinese
 
Dreaming Infrastructure
Dreaming InfrastructureDreaming Infrastructure
Dreaming Infrastructure
 
Big Data, NoSQL, and MongoDB
Big Data, NoSQL, and MongoDBBig Data, NoSQL, and MongoDB
Big Data, NoSQL, and MongoDB
 
大型视频网站单点分析与可用性提升-Qcon2011
大型视频网站单点分析与可用性提升-Qcon2011大型视频网站单点分析与可用性提升-Qcon2011
大型视频网站单点分析与可用性提升-Qcon2011
 
Java 与 云计算
Java 与 云计算Java 与 云计算
Java 与 云计算
 
华为软件定义存储架构分析
华为软件定义存储架构分析华为软件定义存储架构分析
华为软件定义存储架构分析
 
雲端技術的新趨勢
雲端技術的新趨勢雲端技術的新趨勢
雲端技術的新趨勢
 
众行业公司系统架构案例介绍
众行业公司系统架构案例介绍众行业公司系统架构案例介绍
众行业公司系统架构案例介绍
 
设计可扩展的Oracle应用
设计可扩展的Oracle应用设计可扩展的Oracle应用
设计可扩展的Oracle应用
 
深入研究雲端應用程式平台-AppFabric
深入研究雲端應用程式平台-AppFabric深入研究雲端應用程式平台-AppFabric
深入研究雲端應用程式平台-AppFabric
 
Dell
DellDell
Dell
 
基于 MySQL 的B2C电商系统前端数据层架构
基于 MySQL 的B2C电商系统前端数据层架构基于 MySQL 的B2C电商系统前端数据层架构
基于 MySQL 的B2C电商系统前端数据层架构
 
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
 
Etu DW Offload 解放資料倉儲的運算效能
Etu DW Offload 解放資料倉儲的運算效能Etu DW Offload 解放資料倉儲的運算效能
Etu DW Offload 解放資料倉儲的運算效能
 
Emc keynote 1130 1200
Emc keynote 1130 1200Emc keynote 1130 1200
Emc keynote 1130 1200
 
Zh tw introduction_to_hadoop and hdfs
Zh tw introduction_to_hadoop and hdfsZh tw introduction_to_hadoop and hdfs
Zh tw introduction_to_hadoop and hdfs
 
数据访问层开发实践
数据访问层开发实践数据访问层开发实践
数据访问层开发实践
 
Velocity2011分享
Velocity2011分享Velocity2011分享
Velocity2011分享
 

2010 ChinaDatabaseCon "Intorduction to Distributed Infrastructure"

  • 1. Intorduction to Distributed Infrastructure Intorduction to Distributed Infrastructure Guang Feng dudu@hunantv.com April 29, 2010
  • 2. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向
  • 3. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统
  • 4. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统 分布式计算框架
  • 5. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统 分布式计算框架 分布式存储引擎
  • 6. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统 分布式计算框架 分布式存储引擎 数据库会议的一些个人思考
  • 7. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统 分布式计算框架 分布式存储引擎 数据库会议的一些个人思考 Language Tower
  • 8. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统 分布式计算框架 分布式存储引擎 数据库会议的一些个人思考 Language Tower 数据库性能监控与分析
  • 9. Intorduction to Distributed Infrastructure Topic 当今分布式技术主要有以下几个研究方向 分布式文件系统 分布式计算框架 分布式存储引擎 数据库会议的一些个人思考 Language Tower 数据库性能监控与分析 DBA团队建设与职责
  • 10. Intorduction to Distributed Infrastructure Distributed File System 解决的问题
  • 11. Intorduction to Distributed Infrastructure Distributed File System 解决的问题 大容量文件存储
  • 12. Intorduction to Distributed Infrastructure Distributed File System 解决的问题 大容量文件存储 本地文件系统的I/O瓶颈
  • 13. Intorduction to Distributed Infrastructure Distributed File System 解决的问题 大容量文件存储 本地文件系统的I/O瓶颈 灾难备份
  • 14. Intorduction to Distributed Infrastructure Distributed File System 解决的问题 大容量文件存储 本地文件系统的I/O瓶颈 灾难备份 主要产品
  • 15. Intorduction to Distributed Infrastructure Distributed File System 解决的问题 大容量文件存储 本地文件系统的I/O瓶颈 灾难备份 主要产品 面向大文件存储 Google GFS like...
  • 16. Intorduction to Distributed Infrastructure Distributed File System 解决的问题 大容量文件存储 本地文件系统的I/O瓶颈 灾难备份 主要产品 面向大文件存储 Google GFS like... 面向小文件存储 MogileFS
  • 17. Intorduction to Distributed Infrastructure Distributed Computing Framework 概念
  • 18. Intorduction to Distributed Infrastructure Distributed Computing Framework 概念 MapReduce
  • 19. Intorduction to Distributed Infrastructure Distributed Computing Framework 概念 MapReduce 并行处理的是数据,不是任务(计算能力)
  • 20. Intorduction to Distributed Infrastructure Distributed Computing Framework 概念 MapReduce 并行处理的是数据,不是任务(计算能力) 产品
  • 21. Intorduction to Distributed Infrastructure Distributed Computing Framework 概念 MapReduce 并行处理的是数据,不是任务(计算能力) 产品 Hadoop
  • 22. Intorduction to Distributed Infrastructure Distributed Computing Framework 概念 MapReduce 并行处理的是数据,不是任务(计算能力) 产品 Hadoop *Gearman
  • 23. Intorduction to Distributed Infrastructure Distributed Computing Framework Gearman Figure: Gearman Arch
  • 24. Intorduction to Distributed Infrastructure Distributed Computing Framework Gearman work in parallel Figure: Gearman Arch
  • 25. Intorduction to Distributed Infrastructure Distributed Computing Framework Gearman work in parallel background task Figure: Gearman Arch
  • 26. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage
  • 27. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable
  • 28. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable Amazon Dynamo
  • 29. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable Amazon Dynamo Not Only SQL
  • 30. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable Amazon Dynamo Not Only SQL Tokyo Cabinet
  • 31. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable Amazon Dynamo Not Only SQL Tokyo Cabinet Relational Database
  • 32. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable Amazon Dynamo Not Only SQL Tokyo Cabinet Relational Database on-Proxy
  • 33. Intorduction to Distributed Infrastructure Distributed Storage Engine Colunmn-Based Storage Google BigTable Amazon Dynamo Not Only SQL Tokyo Cabinet Relational Database on-Proxy Cluster
  • 34. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势
  • 35. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强
  • 36. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强 每单元(Tracker)中记录统一数据的多个版本
  • 37. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强 每单元(Tracker)中记录统一数据的多个版本 数据聚集性强
  • 38. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强 每单元(Tracker)中记录统一数据的多个版本 数据聚集性强 缺点
  • 39. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强 每单元(Tracker)中记录统一数据的多个版本 数据聚集性强 缺点 不适合频繁CRUD
  • 40. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强 每单元(Tracker)中记录统一数据的多个版本 数据聚集性强 缺点 不适合频繁CRUD 不支持条件查询
  • 41. Intorduction to Distributed Infrastructure Colunmn-Based Storage 优势 动态增加存储列,水平切分伸缩性强 每单元(Tracker)中记录统一数据的多个版本 数据聚集性强 缺点 不适合频繁CRUD 不支持条件查询 不适合OLTP(On-Line Analytical Processing)
  • 42. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性
  • 43. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性 键值对存储
  • 44. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性 键值对存储 insert 0.4s/1M records & search 0.33s/1M records
  • 45. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性 键值对存储 insert 0.4s/1M records & search 0.33s/1M records 支持丰富的语言调用(HTTP协议)
  • 46. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性 键值对存储 insert 0.4s/1M records & search 0.33s/1M records 支持丰富的语言调用(HTTP协议) 不足
  • 47. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性 键值对存储 insert 0.4s/1M records & search 0.33s/1M records 支持丰富的语言调用(HTTP协议) 不足 上亿的数据存储后,性能下降很快
  • 48. Intorduction to Distributed Infrastructure NoSQL:Tokyo Cabinet 特性 键值对存储 insert 0.4s/1M records & search 0.33s/1M records 支持丰富的语言调用(HTTP协议) 不足 上亿的数据存储后,性能下降很快 很难扩展到多机
  • 49. Intorduction to Distributed Infrastructure RDB:Amoeba Proxy Architecture(代理风格)
  • 50. Intorduction to Distributed Infrastructure RDB:Amoeba Proxy Architecture(代理风格) 特性
  • 51. Intorduction to Distributed Infrastructure RDB:Amoeba Proxy Architecture(代理风格) 特性 提供规则数据切分
  • 52. Intorduction to Distributed Infrastructure RDB:Amoeba Proxy Architecture(代理风格) 特性 提供规则数据切分 降低db 与客户端的连接数
  • 53. Intorduction to Distributed Infrastructure RDB:Amoeba Proxy Architecture(代理风格) 特性 提供规则数据切分 降低db 与客户端的连接数 读写分离
  • 54. Intorduction to Distributed Infrastructure RDB:Amoeba Proxy Architecture(代理风格) 特性 提供规则数据切分 降低db 与客户端的连接数 读写分离 未来打算支持异构数据库
  • 55. Intorduction to Distributed Infrastructure RDB:EnterpriseDB Cluster 特性
  • 56. Intorduction to Distributed Infrastructure RDB:EnterpriseDB Cluster 特性 内置连接池,内存共享
  • 57. Intorduction to Distributed Infrastructure RDB:EnterpriseDB Cluster 特性 内置连接池,内存共享 读请求负载均衡
  • 58. Intorduction to Distributed Infrastructure RDB:EnterpriseDB Cluster 特性 内置连接池,内存共享 读请求负载均衡 所有节点并行查询
  • 59. Intorduction to Distributed Infrastructure RDB:EnterpriseDB Cluster 特性 内置连接池,内存共享 读请求负载均衡 所有节点并行查询 查询缓存
  • 60. Intorduction to Distributed Infrastructure Language Tower 分布式基础架构
  • 61. Intorduction to Distributed Infrastructure Language Tower 分布式基础架构 语言无关的通讯协议
  • 62. Intorduction to Distributed Infrastructure Language Tower 分布式基础架构 语言无关的通讯协议 动态脚本语言实现业务逻辑
  • 63. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析
  • 64. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控
  • 65. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控 数据库日志(二进制日志、慢查询)分析
  • 66. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控 数据库日志(二进制日志、慢查询)分析 根据业务特点进行针对优化
  • 67. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控 数据库日志(二进制日志、慢查询)分析 根据业务特点进行针对优化 DBA团队建设与职责
  • 68. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控 数据库日志(二进制日志、慢查询)分析 根据业务特点进行针对优化 DBA团队建设与职责 日常维护数据
  • 69. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控 数据库日志(二进制日志、慢查询)分析 根据业务特点进行针对优化 DBA团队建设与职责 日常维护数据 Top Query的分析与解决
  • 70. Intorduction to Distributed Infrastructure Thinking... 数据库性能监控与分析 结合主机负载与数据库指标进行监控 数据库日志(二进制日志、慢查询)分析 根据业务特点进行针对优化 DBA团队建设与职责 日常维护数据 Top Query的分析与解决 SQL由谁来写?
  • 71. Intorduction to Distributed Infrastructure Thank you! Q&A