A Case of Meta-scheduling for Biological Researches in Cloud Environment

Meta-scheduling is proven to be an essential tool for resource management in cross-domain environments like Grids. This paper takes one step further in combining meta-scheduling with application management tool to present a novel multi-domain cloud environment for biological research. With the scalable architecture and stable performance, Community Scheduler Framework (CSF) is a mature meta-scheduler for traditional Grids. However, as a scheduler, CSF does not support flexible application management. In this paper, we integrate CSF with another useful tool OPAL, which is a toolkit for wrapping scientific application as web services, to take advantage of both grid and cloud resources. We first talk about the integration of CSF4 with Opal to form a user-transparent and configurable scientific cloud platform. Next, some enhancements of CSF4 are presented to improve the usability of the OPAL-CSF system. In addition, we will describe a virtual screening web service in OPAL-CSF system. The virtual screening service compares multiple receptor/ligand protein pairs using Autodock. Performance analysis of this service shows that the integration of Opal and CSF can efficiently facilitate large-scale scientific applications.