In the face of rising data volume and complexity and increased need for self-service, enterprises need an effective business intelligence (BI) reference architecture to utilize BI as a key corporate asset for competitive differentiation.
BI stakeholders — such as project managers, developers, data architects, enterprise architects, database administrators, and data quality specialists — may find the myriad choices and constant influx of new business requirements overwhelming. Forrester's BI reference architecture provides a framework with architectural patterns and building blocks to guide these BI stakeholders in managing BI strategy and architecture.
Enterprise information management (EIM) is complex — from a technical, organizational, and operational standpoint. But to business users, all that complexity is behind the scenes. What they need is BI, an interface to enterprise data — whether it's structured, semistructured, or unstructured. Our June 2011 Global Technology Trends Online Survey showed that BI topped even mobility — the frontrunner in recent years — as the technology most likely to provide business value over the next three years.
BI architecture is never simple, and this is especially true in large, heterogeneous enterprises with global reach and multiple product and service lines. Such enterprises always have more than one enterprise data warehouse (EDW), hundreds of data marts, and several BI platforms. BI reference architecture has to account for agility enablers, such as allowing exceptions to the standards and business user self-service, due to ever-changing business and regulatory requirements. The resulting BI reference architecture may not look pretty and simple, but it's pragmatic and practical.
In this new research - as part of our upcoming BI Playbook - we provide architectural diagrams, explanations and rationalization for the following BI reference architecture components:
- Data sources: where the data comes from
- Data rationalization – mapping apples to oranges. This layer includes
- Virtualized data access
- Extract, transform, load (ETL)
- Business event processing
- Complex event processing (CEP)
- Text and natural language processing (NLP)
- Data quality
- Master data management (MDM)
- Data governance enabling tools
- Direct data source access
- Derived data sources – where a single enterprise data warehouse (EDW) may not be a practical option
- Staging areas
- Operational data store (ODS)
- Data warehouse (DW)
- Data marts
- OLAP cubes
- Analytical data virtualization or semantic layers
- Data usage – what business users touch and feel
- Ad-hoc queries
- Exploration and discovery
- Advanced and predictive analytics
- Process analytics
- Analytical performance management
- Data delivery – where it all ends up
- Office apps
- Other components that span all layers of the BI reference architecture
- Integrated metadata
- Information life-cycle management (ILM)
- Enterprise content management (ECM)
- “BI out of the box” applications
- BI on BI
- Embedded BI / BI services
- Big data
For detailed explanation of each component and diagrams on how it all fits together, please read our latest report
This blog originally appeared at Forrester Research.