Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 9 Adv. DBMS Data Warehouse CSC5301 Review Hachim Haddouti.

Similar presentations


Presentation on theme: "1 9 Adv. DBMS Data Warehouse CSC5301 Review Hachim Haddouti."— Presentation transcript:

1 1 9 Adv. DBMS Data Warehouse CSC5301 Review Hachim Haddouti

2 2 9 Do You Remember? OLTP DSS MD drill down RollUp Slice/dice MOLAP ROLAP Star schema Data mining Data cube Data extraction Fact table

3 3 9 Data Warehouses  “Subject-oriented, integrated, time-variant, nonvolatile collection of data in support of management’s decision-making process” Inmon ( AP = analytical processing is missing) u Used for analysis of existing data u Resolves performance issues suffered by operational RDBMSs and OLTPs

4 4 9 Sizing DW? Mining of mobile phone calls: (Caller, Callee, Time, Duration, Geogr. Location) ~ 100 B/tuple In Germany 10 7 users * 10 calls/(day*user) * 100 B/call = = 10 10 B/day ~ 3*10 12 B/year = 3 TB/year Scanning data at 10 7 B/s takes 3*10 12 /10 7 = 3*10 5 s > 3 days

5 5 9 Data Warehouse Architecture

6 6 9 Data model ER Model u a disaster for querying a huge amount of data (time) u not understandable for users and they can not be navigated usefully by DBMS software. u hard to visualize; many possible connections between tables, u To avoid redundancy MD Model u better performance u Better data organisation u Better visualization u Business queries (why, what if)

7 7 9 Typical DWH Analyses/Queries u What are the consequences of new orders for production capacity w.r. to investment, personnel, maintenance, extra hours,... u Seasonal adaptions, e.g. when to produce how many skis, bikinis, convertibles,... u Influence of external financing on profits

8 8 9 Operations: aggregation slice dice (cube) rollup to coarser level drill down to more detailed level grouping sorting

9 9 9 Data Cube Representation

10 10 9 Steps to build a DWH u Acquisition of data u Data cleansing u Storage u Processing: AP u Maintenance,... Not possible with classical DB-technology alone

11 11 9 On-Line Analytical Processing u OLTP (online transaction processing) for operational data of enterprise, e.g. in relational DBMS, IMS, SAP/R3,... u DSS: Decision Support System to store data/information for strategic management decisions: aggregations, summaries, etc. u Optimized to work with data warehouses u Used to answer questions u Allows users to perceive data as a multidimensional data cube u Data mining

12 12 9 OLTP versus OLAP Thematic focus u OLTP: many small transactions (microscopic view of business processes, individual steps at lowest level, single order, delivery) u OLAP: finances in general, personnel in general,... u OLAP requires integration and unification of many detailed data into big picture u Time orientation u Durability: data extracted once, no updates

13 13 9 Technical Comparison OLTP vs OLAP u OLTP: high rate of updates, several thousand t/s u OLAP: read only transactions, very complex, DWH is loaded at certain time intervals, e.g. after the end of the month, quarter l Compute intensive l Special systems with new access methods, e.g. multidimensional data organization and access methods l Special OLAP systems necessary to offload OLTP systems

14 14 9 ROLAP and MOLAP Solution 1: ROLAP relational online analytical processing, built on top of relational DBS, additional middleware or client front end (star schema) Solution 2: MOLAP: multidimensional online analytical processing u new model u new data organizations u new algorithms u new query languages u new optimization techniques

15 15 9 DW Review degenerate dimension big dimensions hierarchies snow falcking Slowly changing dimensions dirty dimensions Hetegrogeneous prodcuts (core and custom) Factless Fact table


Download ppt "1 9 Adv. DBMS Data Warehouse CSC5301 Review Hachim Haddouti."

Similar presentations


Ads by Google