Empirical validation of cyber-foraging architectural tactics for surrogate provisioning

Abstract

BackgroundCyber-foraging architectural tactics are used to build mobile applications that leverage proximate, intermediate cloud surrogates for computation offload and data staging. Compared to direct access to cloud resources, the use of intermediate surrogates improves system qualities such as response time, energy efficiency, and resilience. However, the state-of-the-art mostly focuses on introducing new architectural tactics rather than quantitatively comparing the existing tactics, which can help software architects and software engineers with new insights on each tactic. AimOur work aims at empirically evaluating the architectural tactics for surrogate provisioning, specifically with respect to resilience and energy efficiency. MethodWe follow a systematic experimentation framework to collect relevant data on Static Surrogate Provisioning and Dynamic Surrogate Provisioning tactics. Our experimentation approach can be reused for validation of other cyber-foraging tactics. We perform statistical analysis to support our hypotheses, as compared to baseline measurements with no cyber-foraging tactics deployed. ResultsOur findings show that Static Surrogate Provisioning tactics provide higher resilience than Dynamic Surrogate Provisioning tactics for runtime environmental changes. Both surrogate provisioning tactics perform with no significant difference with respect to their energy efficiency. We observe that the overhead of the runtime optimization algorithm is similar for both tactic types. ConclusionsThe presented quantitative evidence on the impact of different tactics empowers software architects and software engineers with the ability to make more conscious design decisions. This contribution, as a starting point, emphasizes the use of quantifiable metrics to make better-informed trade-offs between desired quality attributes. Our next step is to focus on the impact of runtime programmable infrastructure on the quality of cyber-foraging systems.