Whether you call it hosted, on-demand, software as a service or cloud computing, the idea of using shared systems managed by someone else is being tested for every conceivable technology function, including BI. We know technologists are famous for pushing good ideas beyond their natural limits, so it's worth asking whether on-demand BI truly makes sense.
In some ways, the fit makes good sense. BI systems are primarily analytical, not operational in the sense of order processing or billing. This means a BI system need not be tightly integrated with the other enterprise processes and can endure unexpected downtime without catastrophic results. Integration and availability are both significant issues for on-demand products.
On-demand BI faces hurdles as well. The most daunting is sheer diversity. Popular on-demand applications for collaboration, content management and sales automation are largely the same from company to company. Not so with BI, where most implementations integrate a unique set of data sources into a custom-tailored structure. This makes standardization difficult and requires personal attention from technical experts, and savings from standardization and automation are precisely how on-demand systems make their money. Without them, on-demand has no particular advantage over conventional approaches.
On-demand BI vendors have addressed these issues in different ways. Their strategies include:
- Predefined sources. Vendors can focus on specific applications. This lets them prebuild connectors to the source data, map their contents into a standard database design and create standard reports. Technically it's a simple approach, no different from the vertical application packages commonly built for conventional BI systems. Nevertheless, it lets vendors provide substantial benefits to their clients.
- Tools to let users do the work for themselves. Common examples are wizard-driven interfaces to set up data integration rules and define reporting requirements. For the vendor, this generally boils down to putting a new interface on an existing tool. However, it places a substantial burden on clients who may lack the time or skill to do the work. Vendors that take this approach must supplement it with staff experts who help clients when needed.
- Automated tools for design and integration. Several vendors have built systems that automatically import client data, analyze it and create appropriate data models and integration processes. Although business and technical experts should review the results before implementation, automated tools do have the potential to provide a reasonable first pass at the system. This gives the experts something to review and the end users something they can work with right away. Automated design is the most technically demanding approach to on-demand BI, but it also tackles the twin issues of variety and labor most directly.
- Specialized analytical databases. Columnar and in-memory database engines are more efficient at analytical processing than conventional products like Oracle, DB2 and SQL Server. This approach lets BI vendors build successful systems without fine-tuning each implementation and lets the system accommodate new data elements and queries without major redesign. Specialized databases complement automated tools nicely by allowing simpler designs (which are easier to automate), providing adequate performance without optimization and allowing simpler adjustments to the automated system's recommendations. Specialized databases also run on less expensive hardware than conventional technology.
- Automated opportunity discovery. This addresses the ultimate roadblock to BI adoption: users don't always know what to do with the results. Automated data evaluation tools should be adapted to mine for significant business information and present the findings to users. Although conventional software and staff experts can do something similar, automated discovery is particularly appreciated for on-demand systems because it proves the system's value almost immediately. In fact, if the vendor has successfully eliminated labor from the rest of the implementation process, it may be able to make automated analytics part of the sales cycle, reducing the buyer's risk.
Can any of these strategies simplify the delivery of BI systems enough to make them a viable on-demand business? I believe the answer is yes. Right now, offerings tied to specific data sources and applications are most likely to succeed because they combine inherent labor savings with the efficiencies of on-demand infrastructure. Longer term, the greatest potential lies with automated tools and analytical databases. These should let developers create an integrated process that requires little effort from the vendor or its clients, and is substantially cheaper and more flexible than conventional alternatives.
As with any technology, on-demand BI will be limited at first. For now, it's worth testing the automated systems to understand what they really can do. Even if the current reality doesn't quite live up to the promises, bear in mind that the systems will grow over time.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access