Online Performance Queries for Architecture-Level Performance Models
Fabian Gorsler · 2013
As modern software systems are subject to increasing dynamics, achieving an acceptable performance becomes a challenge of increasing concern. Thus, during designand run-time, the performance of a software system needs to be analyzed continuously to avoid contention. Model-based performance prediction is an approach to analyze software systems by predicting their behavior and supporting users to draw conclusions. Available approaches for performance prediction are usually based on their own modeling formalism and analysis tools. Users are forced to gain detailed knowledge about these approaches before predictions can be made. To lower these efforts, intermediate modeling approaches simplify the preparation and triggering of performance predictions. However, users still have to work with different tools suffering from integration, providing non-unified interfaces and the lack of interfaces to trigger performance predictions automatically. Our approach is to provide a novel query language capable of expressing queries for questions like “What is the response time of service X?”. Previous shortcomings are addressed by an interface to integrate different tools. The interface is accessible through a unified query language to trigger performance predictions. The design of the query language is based on a classification scheme with an implementation of an extensible architecture aiming to integrate a broad range of tools and third-party extensions. The query language is evaluated by integrating a prominent approach for performance prediction and controls it while conducting a performance analysis. Two additional approaches are conceptually evaluated for integration and showing promising synergies. Overall the results are encouraging and motivate the future development of our query language and the integration of additional tools.