Anda di halaman 1dari 8

Dimensional Data Modelling

Stars, Snowflakes & a Squashed Octopus


Chris Gath Senior Business Intelligence Consultant Siemens IT Solutions & Services

Copyright Siemens AG 2009. All rights reserved.

My background

Chris Gath - Quantitative techniques specialist - 10 years BI and analytics experience Senior BI Consultant Media BI Strategic Manager Finance Decision Support Analyst Manufacturing

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 2 June-09 Siemens Global Media

Todays talk

What is Business Intelligence? Who uses Business Intelligence, why and what are the priorities? How do you go about designing a Business Intelligence solution? What data modelling techniques are required for successful deployment? Summary

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 3 June-09 Siemens Global Media

What is Business Intelligence?

Business Intelligence is a broad category of applications and technologies for gathering, storing, analysing and providing access to data to help enterprise users make better business decisions Search Data Management Business Intelligence is concerned with the ongoing search for a better understanding of the company, of its products and of its customers. Information is explored to help develop new products, measure customer retention, evaluate potential new markets and perform countless other tasks Claudia Imhoff
Confidential / Copyright Siemens AG 2009. All rights reserved.
Page 4 June-09 Siemens Global Media

What is Business Intelligence?

Data Warehousing Inmon A subject orientated, integrated, time variant, nonvolatile collection of data in support of managements decision making process AND/OR?

Decision Support Systems Wikipedia An interactive software-based system intended to help decision makers compile useful information from raw data to identify and solve problems and make decisions.

Kimball A copy of transaction data specifically structured for query and analysis

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 5 June-09 Siemens Global Media

Who uses Business Intelligence, why and what are the priorities?
Used by

Used to
Check performance against goals and targets Compare performance in different time periods Discover and confirm trends and relationships Convey information in a more digestible manner Help advocate a position Provide a report of record Forecast and predict future events

Need to provide 1. Flexible information access 2. Cross functional analysis capability 3. Quality assured information 4. Speed of response 5. Reduced dependence on IT

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 6 June-09 Siemens Global Media

Building solutionsthe design strategy solutions

Inmon

Vs.

Kimball

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 7 June-09 Siemens Global Media

Building solutionsthe design strategy solutions

Kimball

Inmon

Federated Approach

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 8 June-09 Siemens Global Media

Data modelling techniques 1. The Squashed Octopus


DECISION how to reduce production costs to meet corporate objectives QUESTION how does production cost & time compare across genre?
PROGRAMME CATEGORY

Sub Process

PRODUCTION PROCESS

Task

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 9 June-09 Siemens Global Media

Data modelling techniques 1. The Squashed Octopus


DECISION how to reduce production costs to meet corporate objectives QUESTION how does production cost & time compare across genre? QUESTION how much content do production teams reuse in their productions?
PROGRAMME CATEGORY

MEASURES
Sub Process

PRODUCTION PROCESS

Task

Actual Cost of Production Production Process Duration Archive Usage (hh:mm:ss) Programme Duration (hh:mm:ss)

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 10 June-09 Siemens Global Media

Data modelling techniques


Programme Category Dimension

2. The Star Schema


Production Process Dimension
Production_process_id (PK) Process_name Process_description Sub_process_name Sub_process_description Task_name Task_description Process_start_date Process_end_date

Programme_category_id (PK) Genre_name Editorial_category_1_name Editorial_category_2_name Editorial_category_3_name Category_start_date Category_end_date

Time Dimension
Time_id (PK) DD_MM_YYYY Month_number Month_text Short_month Year Calendar_quarter Financial_quarter Level_number

Programme_category (FK) Production_process_startdate (FK) Production_process_enddate (FK) Programme (FK) Production_team (FK) Production_process (FK) Production_cost Production_process_duration Archive_used_duration Programme_duration Production_team_id (PK) Department_name Department_description Tearm_name Team_description Team_type Production_team_start_date Production_team_end_date

Production Process StartDate Production Process EndDate

Production Fact Table

Production Team Dimension

Programme_id (PK) Programme_concept Programme_title Programme_language Rights_expiry-date Delivery_date Linear_airings_available Online_airings_available Programme_start_date Programme_end_date

Programme Dimension

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 11 June-09 Siemens Global Media

Data modelling techniques 3. The Snowflake Schema

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 12 June-09 Siemens Global Media

Data modelling techniques 4. Conformed Dimensions


DECISION how to asses the value of our productions QUESTION whats the relationship between audience value and production cost?

Confidential / Copyright Siemens AG 2009. All rights reserved.


Page 13 June-09 Siemens Global Media

Data modelling techniques

5. Data Warehouse Design Combination of contextual data models and information requirements Select data Add a time element where required Incorporate derived data Determine grain Create data summaries Merge entities
Confidential / Copyright Siemens AG 2009. All rights reserved.
Page 14 June-09 Siemens Global Media

What are the main technical performance considerations?

MOLAP
built for fast query performance summary level pre-processed information best with small data sets with low cardinality

ROLAP
built for optimal cube processing and query flexibility handles large data sets with high cardinality limited by SQL language

HOLAP
combines optimal cube processing with fast query performance ability to drill through to detailed data from summary information combines reduced analysis latency with complex query capability
Confidential / Copyright Siemens AG 2009. All rights reserved.
Page 15 June-09 Siemens Global Media

Summary

Performance
Fast information delivery Efficient data processing Efficient data storage

Integration
Reconcilable Corporate view of performance

Flexibility
Versatile, sustainable, consistent and stable Accommodates multiple sets of user requirements
Confidential / Copyright Siemens AG 2009. All rights reserved.
Page 16 June-09 Siemens Global Media

Anda mungkin juga menyukai