A Pattern Emerges

Before you decide on a data model, you should know your business objective

Another example is the availability of information and network systems, strategic assets for many organizations. Yet, the complex configuration of the information resources and interdependencies across different resources make monitoring and diagnosing problems challenging. Typically, system or network alerts generate from different system resources when an availability problem arises. Most of these alerts, however, are symptoms of a problem that originates from another system that supports the resource generating the alert.

For example, an application inaccessibility alert may be due to a network routing failure alert. You may simultaneously be receiving alerts from other monitoring points in the system and network, as well. Even with sophisticated monitoring tools, correlating these alerts to pinpoint the problem and automatically resolve it through a set of predefined actions is a challenge. But the issue revolves around matching the pattern of systems' expected healthy configuration to that of the system in distress and generating alerts and then taking remedial actions to bring the problem configuration to a healthy state. Using the pattern-matching context to think about system or network monitoring issues provides better analysis and solutions design to these issues.

CUTTING-EDGE DEVELOPMENT

Pattern matching also applies to information systems design and testing processes: Use cases are an integral part of Object Management Group's Universal Modeling Language. (You can think of use cases as primarily prearticulated patterns of user interactions to which the actual system implementation must conform.) Once the system is developed, the use cases are converted into test cases to validate the system. Pattern matching is between the test cases and the actual system behavior. In the future, you can design and build systems with the aid of these use cases so that you not only validate the system implementation but also adjust it to conform to the use case pattern.

A related example is the reusability of existing system modules to solve new requirements. You can match the existing system module's behavior pattern with the new one you plan to develop. This approach ensures that the new system conforms to the same architectural standards as the older ones. Incremental development also reduces development costs by orders of magnitude in contrast to development from scratch. This approach is not only applicable to the system's code development, but also to the technology platform and resource skill base as well. You acquire new technologies that have better integration points with existing technologies and build a resource skill base that best augments existing ones.

Rate This Article Rating: 1 (best) 2 3 4 5 (worst) Comments: Optional e-mail address:

But what if you don't know the pattern at all or the pattern has not yet evolved? Data mining helps you search for unexpected patterns with neural networks, the pattern forms as the system begins to learn and evolve. Both these technologies are valuable and will likely become mainstream technologies in the future.

As you can see, I'm not proposing a new technology paradigm, but rather a shift in the focus and sequence of analyzing business issues and designing information systems. Choosing a priori a particular pattern form or data and function model for the system is a common tendency; instead, you should first define the edge of the pattern match and the actions you need and then work backward and choose the most appropriate pattern form.

Raju Kocharekar [rkocharekar@worldbank.org] is a senior information officer in the Corporate Information Systems unit of the World Bank's Information Solutions Group. He has 20 years of professional experience in IT management and support.