Queen Sofia Palace of the Arts, Valencia, Spain
Features
Feature
Modern architecture
Business integrated insight reinvents the analytic paradigm.
by Dr. Barry Devlin
The data warehouse architecture established in the 1980s has stood us in good stead, but the time has come to look at decision making—as it spans the business process spectrum—in a new light. Business today is so interlinked, its information needs so interdependent, its processes so intertwined, and its reaction times so pressured that only a new architected approach can support it.
This new architecture must cover the entire IT support for the business. It must be fully integrated. And it must stretch beyond “simple” intelligence.
Reset assumptions
What’s needed is a new set of foundational principles that support modern decision making. Foremost, business demands the integration of operational and informational support. The range of processes that combine these tasks and the number of businesspeople who use them continues to grow.
For example, data warehouses are frequently required to ingest Web site clicks in real time so a call center agent can see a customer’s activity today and over the last year while talking with the person. Such seamless processes require a consistent and integrated information base spanning multiple environments.
Another widespread trend is the growth in data types—particularly the volume of unstructured content—that are vital inputs to many decision-making tasks.
These trends have enormous implications for information management and storage. Any modern architecture must feature an expanded, enterprise-wide data model that provides high-level consistency across all information types and allows multiple definitions and subsets of enterprise information to support diverse uses. Furthermore, the model and extended metadata that describe the data must be explicitly considered as part of the same resource in order to ensure deep integration and maximum flexibility. The solution is business integrated insight (BI2).
Business integrated insight
Unlike traditional examples, the BI2 architecture places all data—operational, informational and collaborative, as well as metadata—in a single layer called the business information resource (BIR). (See figure.) This is the foundational layer of the architecture.
The middle layer consists of processes, workflows, tasks, applications, tools and so on: in short, all functionality and processing that runs on the enterprise’s computers. This layer, the business function assembly (BFA), depends on the underlying BIR layer for all information, from creation to ongoing usage.
The top layer, the personal action domain (PAD), represents the intent and behaviors of all users of the system—executives, front-line staff and IT—within the company or external to it. Going beyond physical computing resources, the PAD is the people who run and interact with the business. Understanding the scope of needs, goals and behaviors of users in this domain is necessary to design the lower two layers.
Click to enlarge
Business information resource
Within the BIR, information is defined along three principal axes: timeliness, structure and reliability. The placement of any information nugget—defined as the smallest set of related information of business value, including metadata—on these three lines determines to a large extent the level of management it requires, what processes can interact with it and how people can use it:
- Timeliness defines the time period over which an information nugget exists and is considered valid. It broadly mirrors the life cycle of information from creation through use and either disposal or archival. It is therefore vital to apply critical judgment when you decide which technology is appropriate for any particular base information nugget.
- Structure reflects the ease with which meaning can be discerned in information. A continuum of structure applies to all information. Just as you transform and move information along the timeliness axis in data warehousing, you face decisions about whether and how to transform and move data along the structure axis, usually from unstructured to structured.
- Reliability classifies information according to how much faith can be placed in it. Unmanaged and undependable information from external sources plays an increasingly important role in running a business. As the volume of this type of data grows, its placement on this axis—and the definition of rules and methods for handling different levels of reliability—becomes increasingly crucial.
Business function assembly
The process layer—the BFA—can be characterized along two axes: effect and scope. The BFA functionality is structured as well-defined, callable and largely immutable services that perform meaningful activities at a business level. Services may be built upon lower-level services recursively and linked together into transactions or workflows. They can be inserted, replaced or removed with minimal technical expertise from a workflow in a plug-and-play manner. The scope axis reflects this characteristic structure. Activities and workflows, and their actions and interfaces, are all described by metadata in the BIR.
The effect axis, on the other hand, describes how a particular function affects the business and, by extension, the BIR.
The modern business mandates integrated behavior and near instantaneous access to all information across all business processes.
Creative function produces new business entities or instances—such as a new purchase order or an entirely new type of order. Conditioning function modifies existing entities or instances—updating an order or archiving an old order. The effect of conditioning extends over the entire BIR, as it is the mechanism by which information nuggets are transformed and moved along the axes of the BIR.
Analytical function uses existing information to monitor and understand what is occurring inside and outside the business, forming hypotheses about causes and effects. This function does not modify existing infor-mation but can create new information based solely on it.
Personal action domain
The PAD is the most important and, at the same time, most unexpected layer in the BI2 architecture. It is most important because it characterizes how users of the system behave and what they expect as a result of any action. And it’s most unexpected because it doesn’t represent anything that can be physically instantiated in a computer environment.
The PAD represents a very simplified model of human behavior along two axes: intent and gratification time. Modeling what a particular user intends in a specific moment is a vital step to understanding what type of business function is required in the BFA and what is needed in the BIR to satisfy that intent. Active intent initiates an action, causing something to happen. Thoughtful intent gathers information of any appropriate class for use by analytical function.
The gratification time axis reflects the reality that you can have some things now but will have to wait for others. In a modern business, a growing desire and need for immediate gratification places added emphasis on access to in-flight and live data. However, such data might not be internally consistent, so deferred gratification may have to be endured.
New vision
Clearly, the modern business mandates integrated behavior and near instantaneous access to all information across all business processes. The ability to meet these demands is the difference between success and survival.
BI2 reaches this underlying goal and proposes a comprehensive, but conceptually simple, architecture through which a new vision can be created. Despite a radically different approach to architectural layering and a very broad scope, BI2 nonetheless provides a direct evolutionary path from traditional data warehouse implementations.
Dr. Barry Devlin is among the foremost authorities on business insight and a founder of data warehousing. He is founder and principal of 9sight Consulting.
Photography by Miquel Gonzalez
