Put to the Test: Oracle BI Enterprise EditionBuilt on the federated data access and Web-based architecture of Siebel Analytics, Oracle's latest BI offering is competitive, but the admin and query interfaces could stand a few usability improvements. By Cindi Howson, BIScorecard April 23, 2007
Oracle acquired Siebel Systems in the fall of 2005 and with that acquisition, gained a powerful BI product in Siebel Analytics. Contrary to public perception, this BI product was never restricted to Siebel CRM data, and in fact, boasts one of the broadest data access capabilities of any BI suite vendor. In early 2006, Oracle rebranded Siebel Analytics as Oracle BI Enterprise Edition (EE). For long-time Discoverer or Reports users, these products have been positioned as the Standard Edition, with the vendor declaring ongoing support and product improvements.
This review focuses on the key components of Oracle BI EE and why this will be the vendor's flagship BI product even post Hyperion acquisition
Powerful Business Meta Data Layer A key requirement for any business query tool is a metadata layer that shields users from the underlying complexities of physical table structures. Oracle BI EE delivers this via its "subject areas." Subject areas are built in the Administration Tool, a Windows-client shown at right. In building these subject areas, there are three components:
I found the Administration Tool confusing, hindered by mediocre documentation. Routine tasks are not apparent and can be convoluted. For example, in trying to determine how something in the presentation layer relates to the physical, you must first run a query and then navigate to the item. Meanwhile the relationship between these elements is never displayed visually. The process of defining joins could be simplified. If you do not "import" (a misleading term for reading information from the data dictionary) foreign keys, you must create them when defining the joins. The tool advises against importing foreign keys for performance reasons, but if you don't, no joins are detected even when same column names are the same. The graphical representation of the physical model is not presented as a star schema; it's just a jumble of tables. Creating calculated items (price * quantity, for example) is possible, but poorly documented and buried beneath multiple menus (clicking an ellipse button from within a list of logical column mappings eventually invokes the desired expression editor).
Despite these issues, where the meta data layer is most powerful is in its ability to handle multiple schemas and to connect to multiple data sources that include relational (SQL Server, Teradata, etc.), multidimensional (Microsoft Analysis Services, SAP BW but not Essbase), XML, and even spreadsheets. Of leading BI products, only Cognos 8 and BusinessObjects XI with the Data Federator option provide such flexibility. In addition, Oracle BI EE can automatically handle both aggregate navigation and partitions, something only MicroStrategy can lay claim to and partially BusinessObjects (which handles aggregates but not partitions).
Oracle BI EE has an interesting and unique ability to present its meta data layer as an ODBC data source. In this approach, any end user tool " whether Microsoft Excel or Access or even competitive BI tools " can act as a front end to Oracle BI EE. This technical capability at first seemed to me more like competitive maneuvering on the vendor's part, yet in talking to customers, it makes practical and historical sense: Siebel Systems acquired this technology from nQuire in 2001. nQuire's initial goal was to provide federated data access, without alienating the more established BI vendors or telling customers to throw away their existing BI investments. While such capabilities still exist, I doubt such a happy co-existence of competitive BI tools would be widely used beyond Excel and Access.
Integrated Dashboards
Users access Oracle BI EE through a Web-based portal and are normally presented with a dashboard view (see figure at right). The dashboards are based on individual reports built in Answers, and they nicely present multiple information elements in multiple formats (pivot table or chart). Each element in the dashboard is a portlet that users can further interact with to drill down, slice elements or modify. Whereas some dashboard design tools require IT, the Answers Dashboard Editor is ideal for power users to build dashboards. Dashboards can include a global prompt that affects the individual portlets; for example, in the screen shot at left, both the table and chart can be filtered by year. Using the Oracle Delivers module, the dashboard can also display alerts.
While the dashboards are good, they do not include some advanced visualizations (heat maps, spark lines, small multiples, bullet graphs) and interactivity (such as sliders and push buttons) found in some pure-play dashboards.
For busy executives on the road, these Dashboards can be saved in their entirety to a PDF file (many tools only allow one portlet at a time). If customers buy the optional Disconnected Analytics (not tested), the entire dashboard is available offline with full interactive capabilities.
|
New on the BLOG
Is Gartner's Quadrant the Problem, Or Is It How It's Used?
02. 8.2010
Read more from Cindi Howson >>
Add "customer" to Jimi Hendrix' song title and you have a question central to last week's Clarabridge Customer Connections (C3) conference, Are You Customer Experienced? 02. 5.2010 Read more from Seth Grimes >> Quick Thoughts on Sybase/Aleri 02. 4.2010
Read more from Curt Monash >> Most Popular This Week
Intelligent Enterprise Newsletters
Subscribe Here:
| |||||||||||||||||||||||||||||
|
|






