Resilient hybrid Mobile Ad-hoc Cloud over collaborating heterogeneous nodes

Abstract

The emergence of Mobile Ad-hoc Clouds (MACs) promises more effective and collaborative elastic resource-infinite computing. However, the highly dynamic, mobile, heterogeneous, fractionized, and scattered nature of computing resources coupled with the isolated non-cooperative nature of current resource management systems make it impossible for current virtualization and resource management techniques to guarantee resilient cloud service delivery. In this paper, we present PlanetCloud, our MAC management platform with an intrinsic support for resilient, highly mobile, cooperative, and dynamically-configurable MACs. We use PlanetCloud for the construction and management of resilient hybrid MACs (HMACs) over mobile and stationary computing resources. PlanetCloud comprises a trustworthy fine-grained virtualization layer and a task management layer. PlanetCloud employs the concepts of application virtualization and fractionation using intrinsically-resilient and aware micro virtual machines, or Cells in our terminology, to encapsulate executable applicationfractions. Such employment isolates the running application from the underlying physical resource enabling seamless execution over heterogeneous resources, lightweight load migration, and low cost of failure. Integral to PlanetCloud is resource forecasting and selection mechanism, which provide a MAC with future appropriate resource availability in space and time. Further, these features enable a large set of mobile, heterogeneous, and scattered resources to collaborate through PlanetCloud smart management platforms that seamlessly consolidates such resources into a resilient HMAC. Using analysis and simulation, we evaluate a PlanetCloud-managed resilient HMAC. Results show that PlanetCloud can provision high level of resource availability transparently maintaining the applications' QoS while preventing service disruption even in highly dynamic environments. Additionally, results showed that our approach to minimizing the cost of failure and facilitating easy load migration elevates the resilience of the HMAC to a great extent.