NASA is a leading agency for federal government's cloud computing initiative. NASA's private cloud Nebula has grown to support its missions, however, it has not been best utilized to host a large number of NASA HPC applications due to the lack of automated application provisioning tools. Our deliverables aim to bridge the gap to enable rapid application migration from NASA local data centers to Nebula. NASA scientists are also increasingly embracing public cloud computing platforms, e.g., to parallelize image processing using Amazon EC2. Our deliverables will help NASA users to take advantage of low-cost and high-performing public clouds to rapidly create any-size computing cluster, deploy and execute their applications, and leverage new programming paradigm for better parallelization.
Most supercomputers are still 'luxury' hosted by government agencies, and can only be accessible by outsiders via 'calls-for-proposals'. With better-equipped hardware, cloud computing is emerging as a viable platform for supercomputing. Mid-range HPC applications can be offloaded to public clouds with quick turnaround time. With our deliverables, scientists who are not familiar with cloud computing and its provisioning model can still use it as a supercomputer alternative to quickly validate their models/algorithms, and do not let-go of any great idea. Business with unpredictable capacity requirements can use our deliverables to replicate their applications in the cloud at much larger scale, or for the purpose of disaster recovery of failing applications in local data centers.
More »