SAP BusinessObjects Universe Design Evolution, Intelligent Design, Or Just A Big Mess?
Breakout Description Are you new or intermediate universe designer? Or maybe a project manager overseeing the full lifecycle of a BI project? Perhaps a seasoned Crystal Reports developer investigating the benefits of the Business Objects semantic layer? A well-designed universe is the foundation for a successful business intelligence project and satisfied users. Building this foundation begins long before you click on the Information Design Tool application. A combination of evolution and intelligent design, this presentation describes best practices at each stage of the universe life cycle, including requirements gathering, design, development, testing, deployment, and maintenance. Avoid the big mess and deploy successful implementations now.
My Introduction Dallas Marks is a Principal Technical Architect and Trainer at EV Technologies, an SAP Software Solutions partner and Sybase partner focusing on business intelligence and business analytics. Dallas is an SAP Certified Application Associate and authorized trainer for Web Intelligence, Universe Design, Dashboards, and SAP BusinessObjects BI Platform administration. As a seasoned consultant and speaker, Dallas has worked with SAP BusinessObjects tools since 2003 and presented at the North American conference each year since 2006. Dallas has implemented SAP BusinessObjects solutions for a number of industries, including energy, health care, and manufacturing. He holds a master s degree in Computer Engineering from the University of Cincinnati. Dallas blogs about various business intelligence topics at http://www.dallasmarks.org/.
About EV Technologies EV Technologies is an SAP BusinessObjects solutions firm based in the St. Louis Metro Area SAP Software Solutions Partner SAP Certified Solutions provider SAP BusinessObjects Enterprise Certified Virtual Platform Management for SAP Business Analytics and Business Intelligence Creators of Sherlock, making SAP BusinessObjects better for users and administrators
Diversified Semantic Layer An audio podcast by SAP business intelligence nerds, for SAP business intelligence people that won t call themselves nerds Recorded by a bunch of guys in the social media community Don t miss podcasts both on product news and application, as well as interviews with other SAP Analytics experts in the community Latest episode: Why Should I Care About SAP BW? with special guests SAP Mentors John Appleby and Ethan Jewett and surprise guest SAP EVP Steve Lucas Follow on Twitter at @dslayered http://dslayer.net
Agenda The Origins of the Universe Lifecycle of the Universe For each stage of the lifecycle, we ll examine: Key Tasks Common Pi@alls Best PracBces Next Steps Your QuesBons 6
Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? THE ORIGINS OF THE UNIVERSE 7
What is a Universe? A universe is the seman&c layer that maps everyday terms that describe the business environment to corporate data sources. 8
What is the purpose of the Universe? The universe enables non- technical business users to access and manipulate corporate data without knowledge of SQL or MDX. The universe is the foundabon of user adopbon of a corporate business intelligence system. 9
What is User Adoption? A set of on- going processes and procedures that insure that users are equipped to get the maximum value from your organizabon s BI infrastructure More than just training 10
Why Should You Care About User Adoption? If you build it, they sbll may not come You need a job You re not as good looking as Kevin Costner 11
Three methodologies to create a universe Intelligent Design EvoluBon Entropy 12
Intelligent Design the theory that maaer, the various forms of life, and the world were created by a designing intelligence Source: Merriam Webster dicbonary (hap://www.m- w.com/) 13
Evolution a defintion a process in which the whole universe is a progression of interrelated phenomena Source: Merriam Webster dicbonary (hap://www.m- w.com/) 14
Evolution a better definition a process of conbnuous change from a lower, simpler, or worse to a higher, more complex, or beaer state Source: Merriam Webster dicbonary (hap://www.m- w.com/) 15
Entropy a definition the degradabon of the maaer and energy in the universe to an ulbmate state of inert uniformity - a trend to disorder Source: Merriam Webster dicbonary (hap://www.m- w.com/) 16
Mess a definition Result of entropy - a disordered, unbdy, offensive, or unpleasant state or condibon Source: Merriam Webster dicbonary (hap://www.m- w.com/) Photo Source: Flickr.com 17
My Assertion Some universe design projects are doomed before the InformaBon Design Tool is ever launched 18
My Assertion, cont. and now that so many tools (Web Intelligence, Dashboards, Crystal Reports, Explorer, etc.) ublize the universe, it really needs to be right. 19
Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? LIFECYCLE OF THE UNIVERSE 20
Lifecycle of the Universe Prepare Analyze Plan Implement Test Deploy Maintain Prepare Analyze Plan Implement Test Deploy Maintain 21 Let s discuss Key Tasks Common Pi@alls Best PracBces
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? PREPARATION PHASE 22
Preparation Key Tasks IdenBfy Universe Scope Build a Project Team Adopt Standards Kickoff MeeBng 23
Preparation Common Pitfalls Designer doesn't understand the business Lack of Input/ParBcipaBon from User Community Lack of User AdopBon Strategy and Budget 24
Preparation Best Practices Build user involvement into each project phase IdenBfy Subject Maaer Expert (SME) Create and conbnually refine documented universe design standards Standards, however robust, won t mean anything if there isn t an enforcement mechanism (implementabon and tesbng) 25
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? ANALYSIS PHASE 26
Analysis Key Tasks IdenBfy Candidate Objects Determine Data Model (RelaBonal vs. MulB- Dimensional) Important decisions are made in the analysis phase that are like pouring concrete they will set up and harden. 27
Analysis Common Pitfalls Data model incapable of delivering required performance Data is of poor quality Universe is incomprehensible 28 Too large Poor organizabon of objects Poorly named objects These issues are resolved throughout the lifecycle but should be idenbfied and addressed during this crucial project phase
Analysis Best Practices Just give me everything- I don t have time to give you requirements Let reporbng requirements drive data model and candidate objects in universe No reporbng requirements? See first bullet. 29
Analysis Best Practices, cont. Limit number of objects in universe SAP BusinessObjects recommends no more than 500 objects per universe*, although others recommend even smaller number of around 200 objects** These numbers are guidelines, not absolutes. However, if the universe is too large or inbmidabng, users will not use it. OEM universes may be an excepbon because user requirements are not well known. Keep focus on facilitabng user adopbon, not completeness. * Advanced Universe Design Learner s Guide Revision A, SAP BusinessObjects, 2008. (page 131) ** Howson, Cindy. Business Objects XI: The Complete Reference. McGraw- Hill/Osborne, 2006. (page 93) 30
Analysis Best Practices, cont. Use mulbple universes to cover mulbple subject areas, parbcularly unrelated (and unjoined!) ones Finance Supply Chain DistribuBon 31
Analysis Best Practices, cont. 32 Assign subject maaer expert (SME) to assist in class structure, object naming, and help text, esp. if designers possess insufficient business knowledge SMEs are invaluable in resolving conflicts in corporate business vocabulary and hierarchies Now, regarding the data model
Data Models InformaBon Design Tool supports relabonal, OLAP, and mulb- source database pla@orms Universes can be created on virtually any data model, from highly normalized/transacbonal to star- schema But Image Source: SAP BusinessObjects XI 3.0 Universe Design Learner s Guide 33
Fact: Star Schemas are Better Normalized data models (OLTP) are designed to get data INTO the database efficiently. Star schema data models (OLAP) are designed to get data OUT OF the database efficiently. Performance of transacbonal ERP systems degrades significantly when also supporbng analybc BI funcbons. This is not a limita&on of BusinessObjects. It s simply a fact of business intelligence. 34
Fact: Your Data Isn t as Clean as you Think Source ERP data may not have sufficient quality for detailed analysis Outer joins cannot address all issues and degrade query performance Enterprise InformaBon Management (EIM) tools such as SAP BusinessObjects Data Services not only perform data integrabon to star schemas, but can also address data quality Don t let your project fail because a single, trusted version of the truth doesn t exist. 35
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? PLANNING PHASE 36
Planning Key Tasks Create a Project Plan Plan the SAP BusinessObjects Architecture 37
Planning Common Pitfalls Universe is a single, large delivery rather than mulbple evolubonary deliveries Users not included in every project phase (IT failure) Users not involved in every project phase (Business failure) Failing to define specific tasks and objecbves for user acceptance tesbng (UAT) 38
Planning Best Practices Ready, fire, aim! Determine if universe(s) can be broken into phased, evolubonary deliveries Users begin using solubon faster Users provide feedback for subsequent phases that would not be available from a single delivery Make sure users are involved both on paper and in pracbce EffecBve execubve sponsorship can ensure parbcipabon IT team should build and test environment prior to delivery date, not during 39
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? IMPLEMENTATION PHASE 40
Implementation Key Tasks Schema Design Building the Universe 41
Implementation Common Pitfalls Untrained IT Staff Universe looks like data model, not business model Scope Creep 42
Implementation Best Practices Make sure universe designers are trained Authorized classroom training is very effecbve Aaendees are trained to avoid mistakes that consultants are frequently called in to fix Mentor inexperienced designers with experienced ones Outsource mentoring if no in- house capability An outsourced mentor can bring best pracbces from other organizabons and industries Use automated tools to manage and track issues NOTE: Microsou Word and Excel are great tools, but weren t designed for issue tracking and project management 43
Implementation Best Practices Confirm objects are organized into classes according to the user s conceptual data model, not the physical data model Use help text to assist end users, not IT. Use SAP BusinessObjects InformaBon Steward for impact analysis and data lineage, not the comment fields Insure measure objects have database aggregate funcbons Beware of low universe to report rabo 44
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? TESTING PHASE 45
Testing Key Tasks Quality Assurance User Acceptance 46
Testing Common Pitfalls Lack of robust sample data or true producbon data Poor data quality Time spent on data quality, not universe quality Inadequate user acceptance tesbng Lack of user veto power to delay implementabon Lack of IT veto power to delay implementabon 47
Testing Best Practices IT peer review to insure adherence to best pracbces and standards Insure adequate UAT by key project stakeholders Reduce future report development Bme by insuring universe objects (esp. dates, currencies) are correctly formaaed Use automated tools to manage and track issues 48
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? DEPLOYMENT PHASE 49
Deployment Key Tasks Architecture ProducBon Environment Grant User Access Conduct Training 50
Deployment Common Pitfalls InstallaBon and configurabon issues derail solubon delivery Go- Live is the first Bme users see actual, not test, data Insufficient planning and/or budget to train users 51
Deployment Best Practices As with development, phase deployment to users if possible Per department Hierarchy power users first, then casual users Build producbon environment in tandem with development, so it s not a surprise during deployment 52
Prepare Analyze Plan Implement Test Deploy Maintain Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? MAINTENANCE PHASE 53
Maintenance Key Tasks Universe Maintenance EvoluBon 54
Maintenance Common Pitfalls Inadequate documentabon from last iterabon Insufficient knowledge transfer Limited subject maaer experbse within IT Previous design decisions make maintenance difficult - in extreme cases, starbng over is the best opbon 55
Maintenance Best Practices UBlize SAP BusinessObjects InformaBon Steward for impact analysis, lineage, consistency Use built- in audibng capabilibes of the SAP BusinessObjects BI Pla@orm to monitor usage and rebre unused universes and documents Use automated tools to manage and track issues 56
Universe Design: EvoluBon, Intelligent Design, Or Just A Big Mess? NEXT STEPS 57
Get a Second Opinion An external assessment may be helpful in breaking gridlock and taking the next step Company polibcs frequently prevent reason and common sense from being heard an outside perspecbve can help. 58
Recommended Reading Performance Dashboards: Measuring, Monitoring, and Managing Your Business, Second EdiBon, by Wayne W. Eckerson, Wiley, 2011 Achieving User AdopBon: How to Unlock the Full Value of a Business Intelligence ImplementaBon, by Peter Nobes, Business Objects White Paper, 2005 59
Final Thoughts You might not be able to change the world but you CAN change the universe! 60
More Information Contact: Dallas Marks Email: dallas@evtechnologies.com On the Web: http://evtechnologies.com You Should Follow Me on Twitter: http://twitter.com/dallasmarks
Questions?