The WNoDeS Cloud Virtualization Framework: A Macromolecular Surface Analysis Application Case Study

Abstract

The Grid infrastructures, such as the one provided by the European Grid Infrastructure, represent suitable solutions to achieve cost effective computational power. However, these solutions present to Grid users some challenges, i.e. to create a customized computing environment and its management.For this reason a Cloud-like approach is for certain applications more suitable than a pure Grid one.In this paper, we propose an approach based on WNoDeS- the Worker Nodes on Demand Service - to exploit both the computational and storage power of the Grid and the flexibility of the Cloud paradigm. This virtualization framework makes possible, trough a tight integration with Local Resource Management Systems, a dynamic allocation of virtual resources out of a common resource pool. WNoDes has been installed on some INFN Grid sites, allowing selected users to instantiate aset of customized virtual machines on top of the Grid resources. Running the applications on customized machines simplifies the computing model and reduces the failure rates. We show how it is feasible to smoothly perform a real life molecular surfaces analysis use case instantiating resources through WNoDeS.