DOQS REQUIREMENTS ENGINEERING SERIES

Quality-Based Model Integration

Quality-Based Model Integration improves the verification process of assuring that all analysis models can be properly integrated into a single whole. Even a small project is likely to develop dozens, if not hundreds, of analysis process and data models in support of its requirements. Even if every model is correct, there still is no assurance that all of the models will fit together properly. Proper integration requires active steps on the part of the analyst to assure project success.

This seminar explains how to define, verify, and validate integrated models for information systems:

• Cross-checking techniques provide quality control of all completed analysis models, preventing defects in one area of analysis from being carried into subsequent design activities. Because each technique in this curriculum focuses on a slightly different perspective, individual defects are unlikely to remain hidden because it is unlikely that the same mistake would be introduced multiple times across multiple techniques in both data and process disciplines.
• Checking the connectivity of all of the process models assures that a proper set of process pipelines exists to move all external inputs through the business system to the collection of external outputs. Such a macro-level review assures a high level of system cohesion and testability once constructed.
• Verifying the conservation of data as it moves through the process pipelines assures that no effort is wasted on unnecessary data movement and that no unnecessary delays are incurred waiting for data in the wrong order. In the extreme, conservation analysis results in a reconfiguration of the business relationships among the external agents involved in the project scope.
• Conducting a series of focused and narrow audits helps assure that minority perspectives aren’t neglected in the analysis. Projects often concentrate all of their effort on the analysis of the primary business flow, leaving other areas underrepresented in the analysis, to be eventually neglected in the design.

This seminar supports the broadest range of analysis integration activities. Such integration represents a major risk mitigation investment on the part of the project. Where analysis models are well integrated, this investment is minimal. Where integration is weak, these techniques become costly; but still far less costly than allowing integration defects to leak into design.

Seminar Rationale

All too often, projects create extensive models in support of analysis projects only to learn later that some of those models actually contradict each other. Many projects find the final push to complete analysis so exhausting, in terms of energy and budgets, that no time is allocated or spent trying to assure that the models
describe a solution that will actually work. The result is that models are pushed into design, leaving many remaining analysis defects to be detected and corrected during downstream phases.

These techniques are illustrated in a reactive, audit mode in order to maximize usage across the broadest range of projects. However, for projects supporting all analysis modeling activities, this seminar teaches techniques that should be applied proactively across all modeling activities in order to prevent the defects that result in downstream project failures.

Table of Contents

• INTRODUCTION
• A FRAMEWORK FOR INTEGRATION
• DATA-PROCESS INTERACTION
  • Isolation Analysis
  • Exception Analysis
  • Connectivity Analysis
  • Conservation Analysis
• EXTERNAL VALIDATION
  • Industry Standards & Benchmarks
• TRACABILITY AUDITS
  • Security & Control Audits
  • Encapsulation Audits
  • Voice-of-the-Customer Audits
  • Cost-of-Quality Audits
  • Version & Release Audits
• INTEGRATION EVOLUTION

Seminar Uniqueness

Quality-Based Model Integration teaches specific skills and techniques needed to detect, document, and correct inter-model deficiencies prior to turnover to design. The approaches have been defined in such a way as to be usable on a project, even if that project hasn’t been responsible for producing the analysis models to be integrated, such as with purchased software acquisitions, or project outsourcing.

The most significant defects on all application projects occur during design, with most remaining undetected until late in implementation. The cost to most information systems organizations is unknown and unknowable. Addressing these integration issues prior to design transition assures that the bulk of these costs can be avoided.