Self-tuning of software systems through dynamic quality tradeoff and value-based feedback control loop

Abstract

Abstract. Quality requirements of a software system cannot be optimally met, especially when it is running in an uncertain and changing environment. In principle, a controller at runtime can monitor the change impact on quality requirements of the system, update the expectations and priorities from the environment, and take reasonable actions to improve the overall satisfaction. In practice, however, existing controllers are mostly designed for tuning low-level performance indicators instead of high-level requirements. By maintaining a live goal model to represent runtime requirements and linking the overall satisfaction of quality requirements to an indicator of earned business value, we propose a control-theoretic self-tuning method that can dynamically tune the preferences of different quality requirements, and can autonomously make tradeoff decisions through our preference-based goal reasoning procedure. The reasoning procedure results in an optimal configuration of the variation points by selecting the right alternative of ORdecomposed goals and such a configuration is mapped onto corresponding system architecture reconfigurations. The effectiveness of our self-tuning method is evaluated by earned business value, comparing our results with those obtained using static and ad-hoc methods.

Uncertainty handling in goal-driven self-optimization – Limiting the negative effect on adaptation

“…They are programmed as external RMI interfaces using RMI-IIOP technology so that the infrastructure and the running system can be decoupled, distributed and stand-alone. Notice that this implementation of the framework is a reconstruction of the implementation proposed in Peng et al (2012) by following the MAPE-K architecture, integrating uncertainty handling and adopting quantitative goal reasoning.…”

“…For instance, if the response time is getting too long in online shopping, the preference of the soft goal Response Time Be Reduced should be increased such that textual mode will be chosen to make response time close to its expectation. Detailed descriptions of the preference tuning algorithm can be found in Peng et al (2012).…”

“…With well-understood alternative functionalities and their different contributions to relevant quality requirements, requirements goal models (Mylopoulos et al, 1992), where functional and quality requirements are respectively modeled as hard and soft goals, have been used to reason about the optimal reconfigurations for a running system (Lapouchnian et al, 2005;Wang and Mylopoulos, 2009;Welsh et al, 2011;Peng et al, 2012). A quality attribute is correlated to a soft goal to measure its satisfaction based on the corresponding quality value (Letier and van Lamsweerde, 2004).…”

Self-adaptation through incremental generative model transformations at runtime

“…Specifically, the Analyzer Engine aggregates quality values and context values based on the collected data during runtime monitoring. Using the aggregated and expected quality values, it tunes the expectations of quality requirements through a feedback controller proposed in our earlier work [36]. Expectations ∈ [-1, 1] are used to indicate the expected satisfaction levels of quality requirements, which are different under different situations and thus need to be tuned.…”

“…Baresi et al [6] present FLAGS to facilitate requirementsdriven adaptations. Peng et al [36] propose a requirementsdriven self-tuning approach through dynamic quality tradeoff and value-based feedback loop. Chen et al [10,11] propose requirements-driven approaches for survivability assurance of Web systems and optimization of composite services.…”