An Abstraction Architecture for Business Intelligence

Size: px
Start display at page:

Download "An Abstraction Architecture for Business Intelligence"

Transcription

1 Presented by Aliaksei A. Golubev Basic sources: - Dean Reynolds: Introduction to an Abstraction Architecture for BI (White Paper), in ITtoolbox Business Intelligence

2 Business Intelligence is an abstraction architecture for delivering specific and useful information in the midst of the data-explosion organizations are facing today. Functions corporate weapon to fight fraud, waste, and abuse, empowers decision making at all levels of management, provides real-time notification of business exceptions, advanced reporting and analysis capabilities, ability to play what-if scenarios with operational and historical data to improve tactical and strategic management, facilitates business transparency, compliance enforcement, and data quality improvement. 2 / 14

3 Background in the last century, salesmen starting selling businesses on the idea that they needed computers in order to compete. big boxes and typewriters that punched holes in cards to feed to those big boxes technology improved and IT staffs evolved T i m e l i n e thousand different companies started developing products (and consultants). New hardware, software, and concepts were built. most businesses and vendors decided to focus on specific solutions to specific business problems departments gained autonomy 3 / 14

4 Background 4 / 14

5 Background some vendors began specializing in reporting technologies, ODBC, JDBC, ADO, XML T i m e l i n e lots of magnetic drives to store the source data, the staging data, the reporting data, and the analysis data. Then OLAP technologies and web-based reporting tools. data warehouses Business Intelligence (BI) 5 / 14

6 Abstraction Architecture 6 / 14

7 Extraction Scheduled Extraction using Extraction Tool (PULL) The data is only available at certain times The extraction window is only open at certain times Capturing individual transactions in real-time is not useful Capturing data in real-time is not practical/manageable Your extraction software only allows pulls Asynchronous Extraction using an SOA (PUSH) A need to react to this data immediately A need to evaluate extra-contextual value immediately No acceptable window for bulk loading Source system lends itself to this approach 7 / 14

8 Transformation take data and transactions from source systems and translate them, if necessary, into a form consistent across the enterprise. Dynamic Transformation Contexting Transaction Clearinghouse 8 / 14

9 Loading 9 / 14

10 Loading Enterprise Data Warehouse (EDW) Is a repository of historical data Tightly integrates data Holds the authoritative version of truth Is the centralized source for reporting and analysis On-Line Analytical Processing(OLAP) Relational OLAP (ROLAP) Multi-dimensional OLAP (MOLAP) Hybrid OLAP (HOLAP) 10 / 14

11 Loading Operational Data Store (ODS) Class I ODS real- or nearly realtime Class II ODS more then once a day Class III ODS daily Class IV ODS fed from source systems and DW 11 / 14

12 Presentation data packaging and publishing report generation and analysis delivery and notification 12 / 14

13 Summary eliminate stack of reports, meetings and delays, run a company from laptop on the beach, one plase to go for the answers, managers take decisions in time at any level, data is tightly integrated, common interface, immediate correlation of any two data points in enterprise, let source systems do what they can best, allows forecasting and scenario based planning 13 / 14

14 Thank you!!! 14 / 14