Investigating Oracle EPM System Fusion Edition With Your Data Warehouse Strategy

Presentation By Edward Roske, CEO of interRel Consulting

interRel Consulting focuses mainly on education and works with Hyperion products.  At this year’s OpenWorld, interRel won the 2008 Oracle Titan award for EPM solution of the year.  The presenter, Ed Roske, is CEO of interRel Consulting and author of the book Look Smarter than you are with Hyperion Essbase.

Challenges in a Data Warehousing Environment

Getting data out of a data warehouse can be a lot like drinking from a firehose.  Data warehouses are very imposing.  It’s a little like wondering into a Michael’s warehouse and finding a particular candle.  The challenge is making the data more accessible to the end user.

Users were not happy with IT because there was a growing gap between what the users want and what IT is able to deliver.  Gartner Group termed this the knowledge gap.  The solution is to deliver them the whole platform of information so that users can explore it on their own.

Four Kinds of Users

  • Store manager – Just wants to know what 2000 products sold yesterday so that he knows what to put at the front of the store and what to put at the back of the store.
  • Power users at the corporate headquarters who want to manipulate data using all kinds of involved, ad-hoc SQL queries.
  • Analysts who want a simple, easy-to-use web based tool
  • Executives who just want high-level easy-to-read charts and lists with a dashboard.

Data warehousing gets all the data back into one place, but it is difficult for end users to access.  From an EPM standpoint, do you need a data warehouse to exist?  No, but it makes things a lot easier.  To extend the Michael’s warehouse metaphor, having the data in different data warehousing is like going to a sack warehouse to get a sack, then going to a candle warehouse for a candle.  It makes it more difficult to put everything together and get a unified view.  Dell said that they had a data warehousing solution; it was just in 163 different databases.

These datamarts provide smaller views of the larger data warehouse that optimized for storage and retrieval of total data.  Datamarts are good at comparing aggregate data because they are grouped for a specific subject area

Types of Datamarts

Relational datamarts simply build a summary of the detail data.  Multi-dimensional datamarts aggregate information with drill-down capabilities.

Problem with the two dimensional approach.  Sometimes I do this and sometimes I do that.

Users started creating very complicated spreadsheets.  Cramming multiple dimensions into a spreadsheet.

Essbase was the first one to fit the definition of an OLAP database.  Relational database are good at transactions but not good at analysis.  It is difficult to work with aggregated data in a relational database, but easier in an OLAP database.  End users want to be able to move data around.  That is what OLAP is good at. Essbase provided a highly advanced calculation engine, similar to calculations in an Excel “on steroids”.  It creates multi-dimensions data cubes and allows customers to define their own dimensions.

With OLAP, analysis is fast, flexible, scalable, shared, and secure.  One of the principals of OLAP is that speed should not get any faster or slower regardless of where you are on the scale.
You should be able to change and modify the data in real time and should only be see what applies to your job.

Hyperion wanted to be database-agnostic.  Bring in different structured sources and loads it up into a common data warehouse.  As users began to run what-if scenarios, users wanted the ability to save those plans at the data warehouse level.  With this bi-directionality, Hyperion started to overlap into the Fusion space.

Hyperion realized the value of integrating your data marts with your data warehouse.  Hyperion went on a buying spree and soon had a dozen products that didn’t look anything like each other.
At time it was acquired by Oracle, Hyperion’s catalog had 25 pages of products.  Oracle greatly simplified this down to a half page.  Hyperion is now the “Oracle EPM System, Fusion Edition”.  Oracle also purchased Siebel which it was rebranded as BI Analytics

Evolution of the Nomenclature
OLAP -> DSS -> BI -> EPM

The term BI, or business intelligence, was created by a marketing department at IBM back in 1997.  In the years that followed, IBM spent 100 million dollars marketing the term BI.  The irony is that now everyone knows BI but no one ascribes it to IBM.

EPM stands for enterprise performance management.  EPM systems are used to proactively manage the business, tying operational reporting and analysis to financial plans and reports.  They provide insight into results by giving relevant, timely and actionable information for better business decision making.  Basically, EPM encompasses everything except your operational data.

The Oracle BI Server federates the data with a virtual schema, moving the data “virtually”.  You can build a data warehouse without building a physical data warehouse.  It combines tables across sources (columns + columns) and data in tables sources (rows + rows).  So, you can easily perform transformations and aggregations

Behind the scenes, Hyperion uses Informatica’s Data Integration Manager (DIM) as its ETL Tool.  However, Oracle purchased another ETL tool through its business acquisitions so I expect that Hyperion will be moving in that direction.

How do you become a Hyperion Administrator?

Oracle has roadmaps on Hyperion training.  Generally, there are two weeks of classes on building data marts, followed by one to five day classes on working with different front end applications.

You can also access a copy of Oracle EPM Solution, Fusion Edition through the OTN site.  Available for download under Oracle Enterprise Performance Management System.  Start with the Installation Guide, Foundation Services, and Essbase for the underlying Database

The Hyperion Workspace product is going to be future direction of BI look and feel.  SmartView was Hyperion’s Office add-in.  With it, you can have a table of live data embedded within a Office document.

interRel Consulting offers web training classes every Thursdays at 1pm.  Just go to the interRel Consulting website and check under interRel webcasts for a list of upcoming topics.

ODTUG Kaleidoscope in Monterrey is the best conference to go to if you are interested in EPM.  There, you can get 5 days of training with experts.  It’s an unbeatable value for the price.