Software deployment architecture and quality-of-service in pervasive environments
International workshop on Engineering of software services for pervasive environments: in conjunction with the 6th ESEC/FSE joint meeting
Architecture-driven software mobility in support of QoS requirements
Proceedings of the 1st international workshop on Software architectures and mobility
An architecture-driven software mobility framework
Journal of Systems and Software
Reliability-driven deployment optimization for embedded systems
Journal of Systems and Software
Component deployment optimisation with bayesian learning
Proceedings of the 14th international ACM Sigsoft symposium on Component based software engineering
Population-ACO for the automotive deployment problem
Proceedings of the 13th annual conference on Genetic and evolutionary computation
Entropy-based adaptive range parameter control for evolutionary algorithms
Proceedings of the 15th annual conference on Genetic and evolutionary computation
Hi-index | 0.00 |
The quality of service (QoS) provided by a distributed software system depends on many system parameters, such as network bandwidth, reliability of links, frequencies of software component interactions, etc. A distributed system's allocation of software components to hardware nodes (i.e., deployment architecture) can have a significant impact on its QoS. At the same time, often times there are many deployment architectures that provide the same functionality in large-scale software systems. Furthermore, the impact of deployment architecture on the QoS dimensions (e.g., availability, latency) of the services (functionalities) provisioned by the system could vary. In fact, some QoS dimensions may be conflicting, such that a deployment architecture that improves one QoS dimension, degrades another dimension. In this dissertation, we motivate, present, and evaluate a framework aimed at finding the most appropriate deployment architecture with respect to multiple, and possibly conflicting, QoS dimensions. The framework provides a formal approach to modeling the problem, and a set of generic algorithms that can be tailored and instantiated for improving a system’s deployment architecture. The framework relies on system users’ (desired) degree of satisfaction with QoS improvements to resolve trade-offs between conflicting QoS dimensions. The framework is realized on top of an integrated tool suite, which further aids reusability and cross-evaluation of the solutions. This dissertation is evaluated empirically on a large number of simulated representative scenarios. Various aspects of the framework have also been evaluated on two real distributed systems. The dissertation concludes with several open research questions that will frame our future work.