19. IBM BI 解决方案产品业务系统1业务系统2业务系统3业务系统n……数据仓库管理器/数据库 Warehouse Manager/DB2 UDBDB2 OLAP Server报表工具
QMFDB2 OLAP Server AnalyzerIntelligent Miner for Data其它应用
20. IBM BI体系结构DB2 UDBDB2 UDBDB2 Warehouse Manager
数据仓库管理器Meta DataDB2 OLAP ServerDB2/Warehouse Control Center
OLAP Server App Manager
OLAP Server Analysis Server客户端工具
支持WEB决策支持工具和应用程序DB2 FamilyORACLEInformixSybaseSQL ServerIMS & VSAMFilesData JoinerDB2 Intelligent Miner for Data
数据智能挖掘服务器
27. SQL QueryQuery OptimizerBest Query PlanThreaded CodeCompile -TimeRun - TimeAgentAgentAgentPrefetchersSingle query involves
1 coordinating agent
n sub agents
m prefetchers (shared)
All executing in parallel on available processors
Combination of...
Data parallelism
Each agent works on subset of data
Data dynamically assigned so user not required to partition data
Functional parallelism ("pipelining")
Each agent works on different query function, e.g. scan, sort
Also enables
Parallel Index Create
Parallel Backup and Restore
Allows multiple processes to read or write data to/from the database
Parallel LOAD
Exploitation of multiple processors during load, particularly for parsing/converting/formatting data节点内部并行
28. Parallel Edition - style (shared-nothing)
Data parallelism through hash partitioning
Partitions can be...
Physical on MPP or cluster
Logical on SMPRun - TimeAgentPrefetchersAgentPrefetchersAgentPrefetchersnode 0node 1node nSQL QueryQuery OptimizerBest Query PlanThreaded CodeCompile -Time节点间并行(数据库分区间并行)
29. ................................................................Single Database ViewParallel OptimizerUser
QueryNode
(CPU)Node
(CPU)Node
(CPU)Node
(CPU)Shared-nothing software architecture supports
Independent physical nodes
Separate CPU, memory, and disk
Including SMP nodes
OR
Multiple logical database partitions on single large SMP Server
Interpartition communication is cross memory, not cross network
Data is partitioned across nodes automatically by hashing
Everything operates in parallel
Select
Insert
Update
DeleteBackup/restore
Load
Create index
Reorg充分利用分区数据库的能力
30. Social Insurance NumberNameLocation123-456-789JoeBostonTorontoPartition Key value Hashed to: "8"Vector
Position0123456789101112...Node1231231231231...DB2DB2DB2Partition Map
Determines 'home' for row
Can be adjusted for data skew using the REDISTRIBUTE utilityHash分区和分区映射表
31. Blends best of MPP and SMP style of parallelism
Ideal for SMP clusters
Most flexible hardware support
Leading Edge Query Optimizer!Run - Timenode 0AgentAgentAgentPrefetchersnode 1AgentAgentAgentPrefetchersnode 2AgentAgentAgentPrefetchersSQL QueryQuery OptimizerBest Query PlanThreaded CodeCompile -Time分区内及分区间并行
36. DB2 Connect Enterprise EditionDRDA-Compliant ServerData Replication
Capture
ApplyVisualAge for JavaDB2 ExtendersVisual ExplainDevelopersClient Configuration AssistantUsersDB2 DiscoveryAdministratorCommand Center (GUI CLP) DB2 UDB ServerDB2 UDB ServerGovernorControl Center and Utilities Performance MonitorJob SchedulerPerformance SmartGuideNetwork Configuration SmartGuideAdministration ServerSatellite Administration集成化的图形界面管理工具
37. Control Center
38. Command Center
39. Performance Monitor
40. Performance Smart Guide
41. Index SmartGuide
42. Other Tools Integrated with the DB2 Control CenterDB2 Script Center
Allows users to create and schedule scripts for regular database activities
DB2 Journal
Provides users with a view of activities which have occured in the DBMS
DB2 License Center
Allows users to monitor license compliance
DB2 Information Center
Provides users with the entire DB2 UDB Technical Library online
Server Communications (Network) Configuration Assistant
Automates set up of server for communication with clients
Client Configuration Assistant
Database connection configuration and testing
Can request that DB2 Discovery search network for databases
ODBC administration
DB2 Discovery
Searches for DB2 servers and databases over the network
Returns information required for connection to client
43. DB2家族产品——全面解决方案TCP/IP
IPX/SPX
NETBIOSDOS
WINDOWS
WinNT
Win95
Win98
OS/2
AIX
HP-UX
SCO
SUN Solaris
SNI SINIX
SGI Irix
MAC
Web BrowsersClientsDB2 for OS/400
DB2 for AIX
DB2 for OS/2
DB2 for HP-UX
DB2 for HP-UX 11.0
DB2 for SUN Solaris
DB2 for SINIX
DB2 for NT
DB2 for SCO
DB2 for SCO Unixware 7ServersDB2 Connect
Datajoiner
Net.DataMiddlewareParallel ComplexesDB2 UDB EEE for AIX
DB2 UDB EEE for SUN Solaris
DB2 UDB EEE for Windows NT
DB2 UDB for OS/390
DB2 for OS/400PersonalDB2 for OS/2
DB2 for WinNT
DB2 for Win95
DB2 for Win98
Lotus Approach
Satellite Edition
DB2 EverywhereOracle
Sybase
Informix
SQL Server
IMS
VSAMSourcesTivoli TME-10
Satellite EditionManagementHostsDB2 UDB for OS/390
DB2 for VM and VSE
DB2 for OS/400TCP/IP
SNA
IPX/SPXWANComplete
Solutions
62. 常用数据挖掘算法分为三类Data Mining AlgorithmsNo PredictionPredict
One ThingTime Series MatchingPredict
EverythingAssociationsSequential PatternsDecision TreeRBFClassificationValue PredictionNeuralNeuralClusteringDemographicNeural
63. 常用数据挖掘算法Clustering (Segmentation) - no dependent variable
Demographic Segmentation
Neural Segmentation (Kohonen Map)
Example: Identify common characteristics in a customer data base.
Predictive/Classification Modeling - dependent variable
Nonlinear regression
Decision trees
Neural networks
Radial-basis functions
Example: Predict IBM's stock price tomorrow.
64. 常用数据挖掘算法Link Analysis - transaction data
Basic associations (or dissociation)
Sequential associations (over time)
Example: Identify which features of an insurance policy sell together.
Similar Time Sequence
65. 许多业务问题可以映射到数据挖掘技术
66. Intelligent Miner for Data V6.1
67. IBM Intelligent Miner for DataSequential PatternsAssociationsPredictive ModelingDeviation DetectionClusteringClassificationData Mining KernelsVisualization ToolsIBM DM Pre-Processing LibraryDB2Flat Files数据驱动
发现竞争优势Server端运行于 NT/2000, AIX, SUN Solaris, MVS, AS/400
Client端运行于Win95/NT/2000, AIX, SUN Solaris
68. 分群(Clustering)结果样例
69. 分群(Clustering)结果样例
70. 分类(Classification)结果样例
71. 客户表
公司与客户发生关系的各个方面预测客户可能带来的利润car typecar ageprofessionageamount of claimsvan2.5physicist520van12physician315,000compact1teacher25386sedan0.75teacher551,234compact4secretary39998sports car3car dealer62stat. wagonvicar30sedan1.546compact1.5butcher46预测样例