
A New Approach to Quantifying Architectural Support: QSLS
Choosing the right architecture is critical for any system to meet its quality attribute goals and business objectives. But evaluating architectures has always been a challenge - it often relies on the subjective judgment of experts rather than systematic, quantitative analysis.
A new technique called Quantifying System Levels of Support (QSLS) changes that. QSLS treats architecture and design elements as linguistic concepts and applies AI and mathematical techniques to quantify how well an architecture/design supports key characteristics, quality attributes and business drivers.
Capturing Architecture Semantics
The key insight behind QSLS is that architecture mechanism elements - components, patterns, tactics, etc. - are defined using precise terminology that captures their semantics (note that standards represent a set of mechanisms weighted base on relevance to the standard). Over years of use, the definitions of these elements form a rich ontology representing architectural knowledge. Tools such as OWL, UAF and SysML use the mechanisms to define relationships (qualitative analysis).
QSLS leverages this linguistic information. It uses AI to analyze the semantic relationships between architecture elements based on their definitions. This produces a web of associations, with each link having a correlation score between 0 and 1 indicating its strength.
From Linguistics to Linear Algebra
These semantic relationships are captured in linguistic correlation matrices. Rows and columns represent architecture elements, with cell values indicating the correlation between them.
By applying thresholds to filter out weak correlations, these matrices become focused on the most relevant relationships. Matrix/Vector math (multiplication with weight vectors) can then compute metric vectors identifying the resulting "Level of Support" - a quantitative measure of how well the architecture/design satisfies a given attribute (note that the difference between a perfect score of 1 and the measured Level of Support can be considered RISK of not meeting the attributes full defined capability).
Surfacing Architectural Insights
Calculating support levels provides a powerful tool for architectural evaluation. Alternative architectures can be compared quantitatively to see which better fulfills characteristics, quality attribute scenarios and business drivers.
Low support levels highlight areas where the architecture falls short, allowing targeted improvement. With Books of Knowledge for Architecture, Design and Pre-Implementation, traceability between architecture, design, and implementation becomes quantifiable. This allows for understanding when design or implementation are out of the boundaries defined by architecture and revisions need to take place. These discussions can now include stakeholders to insure alterations are best fitting their needs.
These non-linear calculations allow for direct comparison of different groupings of mechanisms defining different approaches for the architecture/design. This comparison technique allows architects and designers to decide on best approaches. With the computed quality attributes and business drivers, stakeholders can now be better involved in deciding on the direction to fit their needs.
A Systematic, Iterative Approach
QSLS enables a systematic architectural evaluation process:
1. Capture the architecture linguistically (OWL, UAF using QSLS Book of Knowledge)
2. Perform analysis
3. Examine support level metrics
4. Refine the architecture (update mechanisms/standards)
5. Repeat from step 2
This provides an objective, repeatable, and incremental approach to align the architecture with its key objectives defined by stakeholders.
Recorded Demonstration of QSLS Architecture Tool
The Future of Architectural Analysis
QSLS is a new approach for architecture evaluation that replaces subjective judgments with quantitative linguistic analysis. By surfacing insights into how well an architecture supports critical goals, it enables better informed architectural decisions.
Treating architectures as linguistic concepts and leveraging historical knowledge captured in element definitions, QSLS provides a fresh perspective on an old challenge. While an early application technique, enhancing architectural practice is significant.
Comentarios