The very early stage of business intelligence was stacks of format-nonchangeable and plain text mainframe reports. In order to find a number, the business users needed to flip through mountains of paper. The mainframe reporting architecture was very basic (see Diagram 1). So when Excel and Access became an option for reporting, business users were very happy that they could have a mouse-clicking system for reporting. When the volume of reports grew and customization of the reports became an important requirement, people started to rely on Access and Excel as a primary reporting function as long as data could be obtained outside of the mainframe. At that time, every department was building different types of reporting; some data was stored in Access, some was manually entered by the business and sometimes Excel was used to feed an Access database.

 Around the same time as the introduction of Excel and Access, different reporting software introduced reporting server and a parameter-driven reporting technology. People were delighted to produce and look at well-formatted reports with fewer pages. The reporting tools had security features and provided different business units information at different levels (see Figure 1 in PDF below). Then the Internet age came along. When everything started to be done through Internet, the reporting architecture advanced from fat client to zero footprint Web reporting. Internet reporting provides the user the capability of accessing information wherever there is Internet connection, and it also reduces the complexity of the maintenance needs by taking the architecture of zero footprint reporting (Diagram 4 in Figure 1).

Data accumulation started to put a lot burden on the system when both transactions and reporting happened on the same system.  Additionally, an increased number of operational systems, multiple systems for different business lines, and outsourced  hosting required tremendous development effort  to get all the data into a single report. Thus, the enterprise operational data store and data warehousing started to play the role as the integrated enterprise data repository for various reporting.

Business requirements and expectations for reporting functionality have also elevated to a more complex level, especially those power users whose jobs are completely dependent upon the data. Not only do they want the information at their fingertips, but they also want the information anytime, at any place. Not only do they want the information to be presented to them, but they also want to be able to slice and dice the data, aggregate the data at different levels and perform analysis and comparison of the data. Sometimes, reports need to be archived for a lengthy period of time for audit purpose and some Web reports are also used as the source for the downstream systems. As a result, the reporting architecture evolves into a multilayer, multidelivery mode and interrelated infrastructure as what is played in Figure 2.

The tightly connected and interrelated information delivery architecture is only maintainable in the status quo situation. However, with business expansion and higher requirements for data needs in this information-driven world, the new reporting architecture with a semantic layer has been (see example in Figure 3).

In Figure 3, semantics and virtualization play as the foundation for a loosely coupled new reporting architecture. Instead of building reports and a user interface on top of the database layer directly, the virtualization creates a virtual integrated view for users, which allows users to interact with and report on discrete data residing across different systems without a need to refer to each separate system. The illusion created by this virtual data layer in the entire reporting architecture overcomes reporting challenges such as data silos and data islands. Figure 4 basically illustrates a physical implementation of the logic view represented in Diagram 6 in Figure 3. Through data federation server and topology, distributed and disconnected data is virtually and integrally presented to the business.

Based on my experience in information delivery, I have realized that this new reporting architecture changes the conventional way enterprise reports were built and that adoption of this architecture will not only resolve challenges in the report development but also brings many opportunities to enhance the information delivery:

  1. Less development time. Companies are challenged with disparity and distribution of information. Usually, it takes a lengthy period of time to develop new reporting when the underlying data resides across a spectrum of systems. The enterprise data warehouse contains a lot of data, but when new attributes are introduced in the source systems or there are new source systems added, loading all the new attributes and tables into data warehouse still takes valuable time. In this case, data federation can be very helpful. Data federation integrates the data from various heterogeneous sources but still leaves the data in the original sources without creating redundancy. Data federation can reduces the development time significantly for any information delivery project.
  2. Better reporting performance. Report users expect that when they click the button, the report will be displayed on the screen immediately. To address this issue, the conventional reporting architecture creates many pre-built aggregate tables. In the new reporting architecture, data federation servers share a lot of the processing and automatically distribute the workload among them. Also, in the new architecture paradigm, servers (like the reporting servers, federation servers and database servers) will use cache to enhance the reporting performance.
  3. Real-time information delivery. For many operational reports, one day or even one hour latency of data starts to become not acceptable in today’s world. However, most ETL is developed in the batch mode and loading, transforming and aggregating all takes time before information can be delivered to the users. The traditional way to provide near real-time information is through replication, including database replication and disk storage replication. However, database replication will become very costly in the multimaster replication scenario and disk storage replication either has a performance issue or has the possibility of losing the most recent data within the asynchronous replication context. In the new reporting architecture, data federation facilitates the feasibility of real-time information delivery since it doesn’t require all the work involved in the ETL. In the physical implementation of this new architecture, the real-time partition or external real-time database will also be done to separate the real-time data into a different partition or external database.
  4. High availability of information delivery. With all the mobile technology and global service base, information delivery has a higher demand for availability – anywhere, anytime. In the new reporting architecture, with load-balanced data federation layer, the workload can be shared between balancers and within the maintenance window. In addition, since virtualization hides the disparate systems and data, behind the scenes systems can also have fault tolerance and load balancing The systems can also provide the service to a data federation server within their maintenance window so that higher availability and reliability can be achieved.

Single sign-on for all information delivery. With data federation in the new reporting architecture, more reports can and will be produced with more data available to the business. Along with the increasing number of reports and interaction between information and users, the requirement for single sign-on comes to surfaces. So, in the physical implementation of the new architecture, a portal should also be implemented as part of the architecture to minimize the password entries and achieve single sign-on.
The semantic layer solves many challenges of the current information delivery architecture and provides opportunities to offer more data – and more timely data – to the users. It has unveiled a new era for enterprise information delivery.

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

Don't have an account? Register for Free Unlimited Access