Mohamed Ben Belgacem

A hybrid HPC/cloud distributed infrastructure

In this paper, we report on the experimental results of running a large, tightly coupled, distributed multiscale computation over a hybrid High Performance Computing (HPC) infrastructures. We connected EC2 based cloud clusters located in USA to university clusters located in Switzerland. We ran a concurrent multiscale MPI based application on this infrastructure and measured the overhead induced by extending our HPC clusters with EC2 resources. Our results indicate that accommodating some parts of the multiscale computation on cloud resources can lead to low performance without a proper adjustment of CPUs power and workload. However, by enforcing a load-balancing strategy one can benefit from the extra Cloud resources. We connect an EC2-cloud cluster to a university cluster located in Switzerland.We run a distributed multiscale CFD computation over this extended infrastructure.We evaluate and compare the distributed execution to a local execution.We describe our experience of running parallel CFD application on hybrid platforms.Multiscale computation on cloud requires an adjustment of CPUs power and workload.

Distributed Multiscale Computations Using the MAPPER Framework

We present a global overview of the methodology developed within the MAPPER European project to design, implement and run a multiscale simulation on a distributed supercomputing infrastructure. Our goal is to highlight the main steps required when developing an application within this framework. More specifically, we illustrate the proposed approach in the case of hydrology applications. A performance model describing the execution time of the application as a function of its spatial resolution and the hardware performance is proposed. It shows that Distributed Multiscal Computation is beneficial for large scale problems.

Multiscale Computing with the Multiscale Modeling Library and Runtime Environment

We introduce a software tool to simulate multiscale models: the Multiscale Coupling Library and Environment 2 (MUSCLE 2). MUSCLE 2 is a component-based modeling tool inspired by the multiscale modeling and simulation framework, with an easy-to-use API which supports Java, C++, C, and Fortran. We present MUSCLE 2s runtime features, such as its distributed computing capabilities, and its benefits to multiscale modelers. We also describe two multiscale models that use MUSCLE 2 to do distributed multiscale computing: an in-stent restenosis and a canal system model. We conclude that MUSCLE 2 is a notable improvement over the previous version of MUSCLE, and that it allows users to more flexibly deploy simulations of multiscale models, while improving their performance.

Coupling Method for Building a Network of Irrigation Canals on a Distributed Computing Environment

An optimal management of an irrigation network is important to ensure an efficient water supply and to predict critical situations related to natural hazards. We present a multiscale coupling methodology to simulate numerically an entire irrigation canal over a distributed High Performance Computing (HPC) resource. We decompose the network into several segments that are coupled through junctions. Our coupling strategy, based on the concept of Complex Automata (CxA) and the Multiscale Modeling Language (MML), aims at coupling simple 1D model of canal sections with 3D complex ones. Our goal is to build a numerical model that can be run over a distributed grid infrastructure, thus offering a large amount of computing resources. We illustrate our approach by coupling two canal sections in 1D through a gate.

Virtual EZ Grid: A Volunteer Computing Infrastructure for Scientific Medical Applications

This paper presents the Virtual EZ Grid project, based on the XtremWeb-CH XWCH volunteer computing platform. The goal of the project is to introduce a flexible distributed computing system, with i an infrastructure with a non-trivial amount of computing resources from various institutes, ii a stable platform that manages these computing resources and provides advanced interfaces for applications, and iii a set of applications that take benefit of the platform. This paper concentrates on the application support of the new version of XWCH, and describes how two medical applications, MedGIFT and NeuroWeb, utilise it.

A Hybrid Grid/Cloud Distributed Platform: A Case Study

The scene of the computational sciences has considerably changed during the last years. Today, new emerging Desktop grid and Cloud e-infrastructure have a considerable potential to be adopted and used in large scale to exploit thousands of CPUs power to run both scientific and commercial applications. This paper targets scientists and programmers who need to accelerate their scientific research by running their applications on distributed Grid/Cloud infrastructures. We present a hybrid Grid/Cloud platform used to deploy a phylogeny application called MetaPIGA. The aim is to combine the advantages of Grid and Cloud architectures in order to set up a robust, reliable and open platform. We propose two scenarios.

Virtual EZ grid : a volunteer computing infrastructure for scientific medical applications

This paper presents the Virtual EZ Grid project, based on the XtremWeb-CH (XWCH) volunteer computing platform The goal of the project is to introduce a flexible distributed computing system, with (i) a non-trivial number of computing resources infrastructure from various institutes, (ii) a stable platform that manages these computing resources and provides advanced interfaces for applications, and (iii) a set of applications that take benefit of the platform This paper concentrates on the application support of the new version of XWCH, and describes how a medical image application MedGIFT utilises it.

MUSCLE-HPC: A new high performance API to couple multiscale parallel applications

Multiscale, multi-physics applications are central to solve the increasing number of important scientific challenges. Computationally speaking, the difficulty is to combine high performance computing with the need to couple various codes or solvers, each representing a different scale or a different physical process. In this paper, we present MUSCLE-HPC a new HPC implementation of MUSCLE-2, a previously developed Multiscale Coupling Library and Environment. We present its design and implementation and we demonstrate its advantages compared to MUSCLE-2. We conduct a performance comparison through a tightly coupled MPI application use-case. Our results indicate that using MUSCLE-HPC to couple submodels within the same HPC cluster can lead to better computing performance comparable to a native MPI execution and can, thus, reduce the coupling overhead.

Towards a High Level Programming Paradigm to Deploy e-Science Applications with Dynamic Workflows on Large Scale Distributed Systems

This papers targeted scientists and programmers who need to easily develop and run e-science applications on large scale distributed systems. We present a rich programming paradigm and environment used to develop and deploy high performance applications (HPC) on large scale distributed and heterogeneous platforms. We particularly target iterative e-science applications where (i) convergence conditions and number of jobs are not known in advance, (ii) jobs are created on the fly and (iii) jobs could be persistent. We propose two programming paradigms so as to provide intuitive statements enabling an easy writing of HPC e-science applications. Non-expert developers (scientific researchers) can use them to guarantee fast development and efficient deployment of their applications.

A High Level Framework to Develop and Run E-science Applications on Cloud Infrastructures

This paper presents a generic framework used todevelop and deploy high performance applications (HPC) oncloud infrastrcutures. We particularly target iterative e-scienceapplications where (i) convergence conditions and number of jobsare not known in advance, (ii) jobs are created on the fly and(iii) jobs could be persistent. We propose a framework which provides intuitive statementsenabling an easy writing of HPC e-science applications. Non-expert developers (scientific researchers) can use them toguarantee fast development and efficient deployment of their applications.