essbase primer: the basics of a well-designed...
TRANSCRIPT
Essbase Primer: The Basics of a Well-Designed Application
NorCal OAUG / Aug. 26, 2014
Bill ReddyDirector BI & EPM Practice
Analysis Team, Inc.
Doug Pearce Principal Consultant Analysis Team, Inc.
Analysis Team, Inc.
• Essbase & Hyperion EPM Solution Implementation– Business‐centric solutions that improve analysis, management reporting, and planning
– Full project cycle: Requirements, Solution Design and Development, Data Integration, Training
Oracle EssbaseHyperion EPM Suite
Analysis Team, Inc.
• Leading Practices Advisory– Assure strategic alignment of BI and EPM with business objectives
– Develop a prioritized project “roadmap”– Define common business language, KPIs, master data structures (data governance)
– Establish organization, skills and responsibilities to support BI and EPM (organizational effectiveness)
More info at: www.AnalysisTeam.com
• Clients: Mid‐market and Fortune 500
• West coast based & focused
• Growth from repeat business & referrals
Analysis Team, Inc.
Introduction: Bill Reddy
• Over 18 years experience in designing and building Enterprise BI Solutions
• Solution Architect ‐ Full life cycle Essbase & Planning implementations
• Certified Essbase Implementation Specialist• VB, VBA and Java API
Introduction: Doug Pearce
• Over 22 years experience in designing and building Enterprise BI Solutions
• Solution Architect ‐ Full life cycle Essbase & relational implementations
• Essbase Certified• Fuzzy ears and toes – you figure it out
A Well Designed Application is...
• An accurate model of your business• Dependable• Predictable• Optimized but flexible• On supported software
System features understood and driven by the business This Primer is focused on giving you more knowledge to build more impactful applications in order to gain efficiencies, drive revenues and reduce costs.
• Why Essbase– Finance and Beyond – Use Cases
• Principal elements of Essbase– Dimensions, Essbase Outline, Properties and Features– Master Data / Data Integrations (data in, data out)– Calculation Possibilities– Partition “magic”
• Reporting Possibilities• Hallmarks of a Well‐Designed Application• Q&A
Agenda
Why Essbase – Best of Breed• Turns Excel models into system based reporting and
planning powerhouses– Multiple views from the same base data– Calculation rich!– Scalable to thousands of users and millions of hierarchy elements
• Seamlessly integrates data from multiple systems– Aggregate vast volumes of data into a manageable business model
you define
• Drag and drop adhoc with drill‐thru to detail• Awesome prototyping tool, designed for fast development
Finance and Beyond – Use Cases
• Finance– Reporting & Analysis, What If scenarios– GL Consolidations & Eliminations– Planning
• OPEX, Revenue, CAPEX, Employee, Manufacturing Forecasts, Strategic/Long Range Plans
• Other Use Cases– Price‐Volume‐Mix variance analysis– Sales trending and forecasting– Vendor / Supplier analytics– Marketing Effectiveness applications– Reconciling multiple sources of similar data– Driver based models
Dimensions Define the Database
Essbase - a Multidimensional Database
• A data value exists at the intersection of one member from each dimension
• Native Essbase databases start out empty and formless (like a new worksheet)– Additional Oracle products include prebuilt dimensions (off topic for today)
• Dimensions:– Has Hierarchies– Can be very ordered and deep or very flat
A Five-Dimensional Model
Dimensions with Hierarchies
Upper levels in hierarchies define summaries (totals) for reporting and analysis.
Dimensions – less Hierarchical
.Accounts speaks to the type of data in the cube
** Scenario hierarchy suggests the business processes it supports **
Principal Elements of Essbase• 1 Database ‐> 1 Outline
– 1 Outline ‐> many Dimensions • 1 Dimension ‐> many Hierarchies
– 1 Hierarchy ‐> many “Members”
• Members:– Basic element in Essbase (i.e. Jan)– Consolidate (+, ‐, /, *, etc)– Primary name vs aliases (max of 32)– Uniqueness test– Formulas– Stored, Shared, Dynamic, Label‐
Only– Extras: Special Types, User Defined
Attributes (UDA’s)
Alternate Hierarchies:– Alternate roll‐ups– Attribute Dimensions
Dynamic Time Series– First/Last/Avg.– Skip Missing / zero
Outline Properties and Typed Measures
Numeric (Default)TextDate
Member Properties
Shared Member roll-ups
Alternate “Diet” hierarchy that doesn’t add up
Attribute Dimensions
Colas Root Beer Cream Soda Fruit Soda Diet Drinks Product Bottle CanPkg Type
Dynamic Time Series & Time Balance
Principal Elements of Essbase (Cont.)
• Backend elements at a glance:– Substitution Variables
– MaxL Automation Scripts– Report Scripts
Essbase Outline Management• Keep your hierarchies out of trouble and in sync with related reporting systems– Use prefix / suffix– Consistent naming convention– Consider adding a test and approval process so changes are well
understood and “safe”• Review with teams
• The “Dreaded” Re‐Org– Easy to do in Essbase– Historical data is going to move also
• Maybe create alternate hierarchy to preserve history
– Can break reports, calculations, Partitions, security filters, dimension builds and data loads (less likely).• Maybe practice on a test cube
Data Integration• Loading data into Essbase
– Via SmartView Submit Data from Excel– Essbase Load Rules (with text files or SQL loading)
– ODI, Studio, FDMEE– Partitions or XREF or XWRITE between cubes
Data Integration• What about hierarchies?
– Mostly developer focused tools– EAS Console: manual changes (sometimes used by the business)– Essbase Load Rules (text files or SQL)– Essbase Studio, ODI, EPMA, DRM, Outline Load Utility (for Planning)
• Drill‐Thru– Essbase Studio– FDMEE
Data Integrations
• “System” ways to get data out of Essbase– Of course SmartView is the user’s default method– Essbase Report Scripts– Full or partial “Export” of a cube– XREF or XWRITE between cubes– ODI– For the really brave, the Essbase Grid API (available in C or Java)
Calculation Possibilities
• Genius design principle:– “If you can model a calc in Excel, it can be done in Essbase.”
• Allocations, ratios, variances (fav / unfav)• Actualization of Forecasts• Price‐Volume‐Mix• Driver‐based calculations• Financial Statements (P&L, B/S, Cash Flow)• Currency Conversion (to and from)• Varying time dimensions and 52/53 week years (Retail)
Calculation Possibilities
• Calc ahead of time or when running a retrieval?• A more complex scenario:
– Sales and Shipments across all stores for next week– Allocated down to store level based on 13 month actual trend and percent of fleet
– Allocated to hour and factor by each store’s size (Sq. Ft and # of Registers, etc.) and staffing
– Deriving pay rates to calculate payroll expenses by GL account
– And the ability to override calculated results– Calc run from SmartView; data controlled by the business
Partition “Magic”
• Sometimes linking multiple cubes together is greater than the sum of the parts
• Transparent vs. Replicated (Live or Copy)• Allows using the calculation power of ASO and BSO (cube teamwork!)
• Focused Budget cube backed by a ginormous Actuals cube
• Join multiple cubes together seamlessly• Hierarchy sync important but 100% needed?
Transparent or Replicated
Partition Potential
2014 Actual & Budget Product
Family by month
2011 - 2013 Actual
SKU by Day
2014 Actual
SKU by Day
2015 Budget Product
Family by month
There is no magic btw -
LOL
Reporting Possibilities• SmartView
– Adhoc Essbase / Planning– Multiple sheets, multiple ranges– Smart Slices– Planning input forms and offline usage– Also runs reports from HFM, HFR, Web Analysis, HPCM, even OBIEE– Enhance with VBA
• Financial Reports: highly formatted reports with distribution
• Web Analysis• OBIEE: future direction from Oracle
– Has mobile apps
• Planning: forms and adhoc grids– Also has a new mobile app
• 3rd party tools
Hallmarks of a Well-Designed Application• Disclaimer: the following may stump your developers• Dimensions and hierarchies match my reporting needs• A calc or refresh gets the right results every time• The raw data is separated from calculated data it’s auditable• User inputs are easily identified and available for changes• A good mix of pre‐calculated vs dynamic (report time) metrics• Only has the necessary minimum dimensions
– Split a cube into multiple cubes
• A cube that has results that are consistent with my Excel models
• Database is optimized to produce quick results for all standard reporting
Hallmarks of a Well-Designed Application• Test / Automate / Support: questions for your IT folks– What are the specific support steps for changing a month/QTR/year in the system?
– How do users know what is ready to use and when?• Similarly, is our system aware when it’s sources are ready so everything works together?
– How can I quickly test the system when you make changes?
– Can the automation jobs run some tasks in parallel so it all finishes faster?
– Do we have multiple backups that we can use to easily restore our hierarchies and/or data?• Super‐Genius: How sound is the restore procedure?
Business Problem Specific Considerations
• Complex logic but small number of users• Hundreds or thousands of users running reports (esp. Monday morning)
• Global offices• Manual hierarchy updates by the business vs. automated vs. both?
• Simple security or detailed read vs. write• 3rd party tools• Automation sensation
Direct Questions Deserve Honest Answers
• Remember– Do not accept “No”– Do not accept “Slow”– Challenge everything– Think out of the box
Q&A
Contact Information:
Doug Pearce Bill ReddyPrincipal Consultant Director, BI & [email protected] [email protected]
www.AnalysisTeam.com
Stop by our table in the vendor area