You are on page 1of 2

Twitter Linkedin RSS Newsletter IOS App The Android Application

WhitePapers Google +1 Google Currents Facebook

Enhancing business with technology – in association with eweek.com
HOME NEWS TOOLS GALLERIES OPINION EVENTS SECURITY IT LIFE CLOUD SERVERS WEBCASTS SMB VIDEO MOBILE QUIZ GREEN POLLS BIG DATA DATA CENTRE AWARDS APPLE GOOGLE CES 2014 DOWNLOADS

WHITE PAPERS

IT JOBS

TECH CLUB

PARTNER ZONE

TECH SUCCESS AWARDS
5 Myths of Dimensional Modelling
Data warehouses should be fast and accessible. Data modelling is the way to ensure this, says Ralph Kimball
On February 12, 2014 by Ralph Kim ball and Margy Ross 0

Click here to discover the 2013 winners.

SURVEY: HOW MOBILE ARE YOU?
Is your organisation fully embracing 4G and BYOD? Take the survey for your chance to win your free Mobile World Congress PASS.

Dimensional modelling is a technique for data warehouse design that has become popular because it is good at handling end-user queries. Despite its widespread acceptance, however, some misperceptions persist in the industry. These false assertions are a distraction, especially when you want to align your team around common best practices. If folks in your organisation continually lob criticisms about dimensional modeling, their perceptions may be clouded by some common misunderstandings. This list of myths may help: it is extracted from The Data Warehouse Toolkit – The Definitive Guide To Dimensional Modelling by Ralph Kimball and Margy Ross

INFOGRAPHIC: BIG DATA IN A NUTSHELL
Confused about Big Data? TechWeekEurope and IBM break it down for you.

Myth 1: Dimensional Models are only for summary data
This first myth is frequently the root cause of ill-designed dimensional models. Because you can’t possibly predict all the questions asked by business users, you need to provide them with queryable access to the most detailed data so they can roll it up based on the business question. Data at the lowest level of detail is practically impervious to surprises or changes. Summary data should complement the granular detail solely to provide improved performance for common queries, but not replace the details. A related corollary to this first myth is that only a limited amount of historical data should be stored in dimensional structures. Nothing about a dimensional model prohibits storing substantial history. The amount of history available in dimensional models must only be driven by the business’s requirements.

IT LIFE SERIES
Who is behind IT? We want to find out what the IT decision makers are doing and what makes them tick. If you want to take part in our new IT Life series, drop us an email: info(at)netmediaeurope.com.

NEWSLETTERS

Myth 2: Dimensional Models are departmental, not enterprise
Rather than drawing boundaries based on organisational departments, dimensional models should be organised around business processes, such as orders, invoices, and service calls. Multiple business functions often want to analyse the same metrics resulting from a single business process. Multiple extracts of the same source data that create multiple, inconsistent analytic databases should be avoided.

Myth 3: Dimensional Models are not scalable
Dimensional models are extremely scalable. Fact tables frequently have billions of rows; fact tables containing two trillion rows have been reported. The database vendors have wholeheartedly embraced DW/BI and continue to incorporate capabilities into their products to optimise dimensional models’ scalability and performance. Both normalised and dimensional models contain the same information and data relationships; the logical content is identical. Every data relationship expressed in one model can be accurately expressed in the other. Both normalized and dimensional models can answer exactly the same questions, albeit with varying difficulty.

QUIZ
Last quizzes : Lenovo Fibre For Broadband Big Data Edward Snowden And The NSA Highlights of CES 2014 View more quizzes

Myth 4: Dimensional Models are Only for Predictable Usage
Dimensional models should not be designed by focusing on predefined reports or analyses; the design should center on measurement processes. Obviously, it’s important to consider the BI application’s filtering and labeling requirements. But you shouldn’t design for a top ten list of reports in a vacuum because this list is bound to change, making the dimensional model a moving target. The key is to focus on the organisation’s measurement

but you can’t dodge the effort.events that are typically stable. London Business Analyst – Software – Start up . Part of NetMediaEurope. On the contrary. Try our Big Data Quiz! Techweekeurope for mobile devices Vote View Results IT-JOBS HELPDESK TEAM LEADER – SUPPORT TEAM LEADER – FIELD ENGINEER . InTheSpotLight-cloud. As the architect Mies van der Rohe is credited with saying. The correct starting point for your dimensional models is to express data at the lowest detail possible for maximum flexibility and extensibility. NET. dimensional structures are extremely flexible and adaptive to change. and his Kimb all methodology. 2014 by Michael Moore 0 Rackspace CEO Retires As Firm Looks Towards Hybrid Cloud Silent Circle And Apple Kills The Only Geeksphone Create iOS-Compatible Privacy-Focused Bitcoin Trading App Blackphone Canadian Scientists Claim A Major Breakthrough In Quantum Computing Michael Dell Announces “World’s Largest Startup” At Dell World Rackspace CEO Retires As Firm Looks Towards Hybrid Cloud On February 11. Chief Enterprise Applications Architect (Cloud. persistent master data in the ETL system and then reused across dimensional models to enable data integration and ensure semantic consistency. You can’t hold dimensional modeling responsible for organisations’ failure to embrace one of its fundamental tenets. The secret to query flexibility is building fact tables at the most granular level. RECOMMENDED READING Jolla Releases Source Code For Sailfish Browser On February 11. Delivering anything less in your dimensional models undermines the foundation necessary for robust business intelligence.. Data integration depends on standardised labels. About NetMediaEurope .Stockport at Cheshire Datasystem s Lim ited. TOGAF) at . Dimensional models that deliver only summary data are bound to be problematic. data warehousing. POLLS What do you want to see at Mobile World Congress 2014? A Nokia Android phone James Corden Lenovo's Motorola plans HTC One 2 Samsung Galaxy S5 Tizen Yet another smartwatch Myth 5: Dimensional Models can’t be integrated Dimensional models most certainly can be integrated if they conform to the enterprise data warehouse bus architecture. users run into analytic brick walls when they try to drill down into details not available in the summary tables. you’ll likely pre-summarise the data. Wiley. Ralph Kimball. London Internet Author / Website Developer . Developers also run into brick walls because they can’t easily accommodate new dimensions. or dimensional modelling. 2014 by Steve McCaskill 0 LAST COMMENT 0 replies to 5 Myths of Dimensional Modelling Add a comment Safer Internet Day: Neelie Kroes Says Parents Must Protect Children Online On February 11. is a standard in decision support. A related corollary is that dimensional models aren’t responsive to changing business needs.NET Developer/Programmer/Software Engineer. which can be fatal in the long run. Ralph Kimb all is one of the pioneers of data warehousing. It is hard work to reach organisational consensus and then implement the corresponding ETL rules. spotlight. Conformed dimensions are built and maintained as centralised. MVC at . when you pre-suppose the business question. more Jobs . “God is in the details. 2014 by Steve McCaskill 0 PayPal President Has Credit Card Hacked On February 11. NEWS Email 1 Tw eet 1 Like Sign Up to see w hat your friends like.” Delivering dimensional models populated with the most detailed data possible ensures maximum flexibility and extensibility.London at .. 2014 by Max Sm olaks 0 TAGS Android Apple BlackBerry broadband Cloud Infrastructure data centre Facebook Featured Google Government IT Green IT HP microsoft mobile Mobile & Wireless Networking Open Source privacy samsung secure-it internet iPhone Legal Managem ent smartphone smartphones SMB Software and Apps spotlight Storage tablet Window s Security servers Categories Products News Comment Interview Knowledge Centre Events Site About us Contact us Privacy Policy Sales and Advertising Terms and Conditions NetMediaEurope UK TechWeekEurope ChannelBiz Downloads ITweb NetMediaEurope UK France Germany Italy Spain NetMediaEurope © Copyright 2014 All rights reserved. regardless of whether you’re populating normalised or dimensional models. C#. Remember. because of their symmetry. Share 1 Viber Denies Acquisition Rumours On February 11. values.Permanent . Presentation area databases that don’t adhere to the bus architecture with shared conformed dimensions lead to standalone solutions. Stockport . This is an extract from The Data Warehouse Toolkit – The Definitive Guide to Dimensional Modelling b y Ralph Kimb all and Margy Ross.. unlike analyses that are constantly evolving. attributes. or facts with these prematurely summarised tables. 2014 by Max Sm olaks 0 Tags : Big Data. and definitions.London at .