Rapport: Robust Application Porting for HPC in the Cloud.

Lead Research Organisation: Imperial College London
Department Name: Dept of Computing

Abstract

We propose to examine the feasibility of Cloud Computing for Physics and Bionformatics e-Science by porting a set of scientific applications in these domains to a variety of Cloud platforms. This work will be carried out in conjunction with application scientists in Bioinformatics and Physics. The applications that have been selected for piloting on Cloud platforms have been chosen because they present challenges in data volume, processing requirements or close-coupled parallel execution that are typical of most scientific HPC software. Project outputs will be a study of HPC Cloud feasibility which will compare performance of the selected software on a variety of Cloud platforms, seek to determine best practice for Cloud deployment of HPC software both generally and in the chosen domains, and make general recommendations about the use of Clouds for scientific HPC software within an institutional context.

Planned Impact

We outline below the benefits of a better understanding of Cloud Computing for scientific HPC software for five main categories of beneficiaries: Researchers, Research Institutions, the Research Community, Industry, and the general Public. * Researchers ** Application scientists will gain access to more resources. ** Cloud computing will substantially lower the bar to entry, thus making it possible for more researchers to use HPC software. ** Application scientists will also have a broader range of hardware and systems at their disposal. This is in contrast to the situation which currently prevails, in which researchers are must choose between the machines made available at their institution or the grid. ** Wider range of kit/systems could lead to greater flexibility in design. ** Cloud computing increases portability of software across resources. * Research Institutions ** Institutions will be able to support their research communities more flexibly. ** Current spending models for HPC equipment following a peak provisioning model. As a result, resources purchased are rarely fully utilised. Cloud computing has the potential to maximise resource use, leading to overall efficiency. ** Best-practice knowledge will result in more informed planning. * Research Community and Funding Bodies (Funding Councils) ** Best-practice results will guide planning for FCs and Institutions. ** More informed planning for long-term programs in IT. * Industry ** Scientific Cloud has the potential to stimulate further investment in Cloud in industry, as universities move to cloud. * General Public ** Wider use of e-Science has the potential to increase the impact of computer-based scientific research on public sectors such as health, transport and education. ** Use of utility computing by researchers will make networking and connectivity across sectors easier

Publications

10 25 50
publication icon
Cohen J (2013) RAPPORT: running scientific high-performance computing applications on the cloud. in Philosophical transactions. Series A, Mathematical, physical, and engineering sciences

 
Description RAPPORT demonstrated how a range of applications, particularly from Physics and Bioinformatics, could be successfully ported to Infrastructure as a Service Cloud environments. It was shown that important applications from these areas could be ported to Cloud environments without excessive effort and that performance was maintained relative to alterative cluster or Grid environments. This opens the way to more accessible Cloud facilities being used for applications of this type.
Exploitation Route The technology is generic, therefore as applicable in a non-academic context as an academic one. Indeed, availability of public/commercial Clouds means that applications of the type studied would become accessible to public or commercial groups who do not currently have access to HPC resources. Availability of Cloud resources, either public/commercial. e.g. Amazon EC2, or private, group/institutional.
Sectors Chemicals,Digital/Communication/Information Technologies (including Software),Healthcare,Pharmaceuticals and Medical Biotechnology

URL http://cloudresearch.jiscinvolve.org/wp/2011/03/08/rapport-robust-application-porting-for-hpc-in-the-cloud/
 
Description RAPPORT demonstrated that Infrastructure as a Service Cloud environments could be successfully used for HPC-style applications from Physics and Bioinformatics. These applications where typical of a range of HPC applications. Therefore RAPPORT demonstrated that Cloud environments could be used for this class of applications. This considerably widens the accessibility of this HPC technology to a wide variety of users, academic, commercial and public, who currently do not have access to appropriate resources. Demonstration that Cloud environments such as Amazon EC2 could be used for range of HPC applications. Beneficiaries: Society in general Contribution Method: Demonstration of feasibility.
First Year Of Impact 2015
Sector Digital/Communication/Information Technologies (including Software),Pharmaceuticals and Medical Biotechnology
Impact Types Cultural,Societal,Economic

 
Description EPSRC
Amount £483,099 (GBP)
Funding ID EP/I030239/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 01/2011 
End 09/2013