VECFlex: Reconfigurability and Scalability for Trustworthy Volunteer Edge-Cloud supporting Data-intensive Scientific Computing

Although ‘‘volunteer edge-cloud’’ (VEC) computing has emerged as a new paradigm for scientific computing in recent times, wider adoption is being hindered due to the abundance of heterogeneous volunteer resources and lack of trust in their ability to satisfy workflows’ performance and security requirements. In this paper, we propose the VECFlex, a flexible resource management framework that can reconFigure resource security policies to meet workflow requirements and efficiently allocate resources from a large pool with disparate configurations. It employs Reinforcement Learning (RL)-driven resource behavioral modeling to rank the trustworthiness of resources in terms of their availability, flexibility of applying security policies, and consistency of task execution times. VECFlex also applies a security-based modified Particle Swarm optimization (PSO) scheduler to allocate optimal resources to workflow tasks from a large pool of disparate resources. We evaluate the performance of VECFlex using an AWS testbed that demonstrates VECFlex’s ability to fully satisfy workflows’ security requirements and $\sim$2x improvement in workflow execution latency.