Sunday, June 7, 2015

BI DWH OLTP DRAFT

Introduction to Basic BI Concepts
What is Business Intelligence
Definitions
•Howard Dresner’s definition - business intelligence is an umbrella term to describe ‘concepts and methods to improve business decision making by using fact-based support systems’
•Business intelligence (BI) is the set of techniques and tools that transform raw data into meaningful and useful information for the purpose of business analysis

What can we understand from the Definitions?
•concepts and methods / techniques and tools
–BI is more than a set of tools and technologies
•business decision / business analysis
–BI is driven by the need of business to take informed decisions
•fact-based / raw data  useful information
–Analysis is based on raw data / facts

What is data / “raw facts”
Data / Raw Facts
•Facts generated by operational workflow of a business process
–This training has generated data
»Topic, trainer name, date, duration, info about participants …
•Facts generated by non-business events
–Weather information
–Information gathered by traffic sensor
•Can be digital, paper, structured, unstructured, semi-structured
•Can be recorded somewhere or just in a person’s head


Where are raw facts / data created?

OLTP System
•An application or modules of application that automates a specific area of business operations workflow, such as pre-sales, sales, procurement, r&d, accounting etc.
•Implements current business rules that govern operations
•Focuses on managing current data slice rather than historical data
–historical data may be purged / archived to improve database performance or to reduce data storage costs or implement new business rules that earlier data does not comply with
•Focuses on managing large volumes of simultaneous transactions, involving low volumes of data per transaction
•May support operational reporting, but full-fledged analysis is not the core purpose
•Data repository usually RDBMS (exceptions are mainframe / COBOL applications etc. & non-transaction data such as ATM logs, network signals, traffic signal data, etc.). Data model is usually normalized


0 comments :

Post a Comment