Simplifying the Use of Clouds for Scientific Computing with Everest

Cloud computing has emerged as a new paradigm for on-demand access to a wast pool of computing resources that provides an alternative to using on-premises resources. This paper discusses the challenges related to using the cloud computing infrastructures for scientific computing. An approach based on Everest platform addressing these challenges is presented along with the prototype integration of Everest with Google Compute Engine. The proposed integration enables Everest users to seamlessly provision and use cloud-based computing resources for running different types of workloads including HPC and HTC applications. In contrast to other efforts, the presented approach also supports building and sharing domain-specific web services that automate execution of applications on dynamically provisioned cloud resources or hybrid infrastructures.