This paper formulates a pattern search method for nonlinear programming. A distributedparallel algorithm is given based on the pattern search method by distributing the direction to every processor. Promising experim...
详细信息
Fulfilling a service request in highly dynamic service environments may require deploying a service. Therefore, the effectiveness of service deployment systems affects initial service response times. On Infrastructure...
详细信息
Fulfilling a service request in highly dynamic service environments may require deploying a service. Therefore, the effectiveness of service deployment systems affects initial service response times. On Infrastructure as a Service (IaaS) cloud systems deployable services are encapsulated in virtual appliances. Services are deployed by instantiating virtual machines with their virtual appliances. The virtual machine instantiation process is highly dependent on the size and availability of the virtual appliance that is maintained by service developers. This article proposes an automated virtual appliance creation service that aids the service developers to create efficiently deployable virtual appliances - in former systems this task was carried out manually by the developer. We present an algorithm that decomposes these appliances in order to replicate the common virtual appliance parts in IaaS systems. These parts are used to reduce the deployment time of the service by rebuilding the virtual appliance of the service on the deployment target site. With the prototype implementation of the proposed algorithms we demonstrate the decomposition and appliance rebuilding algorithms on a complex web service. (c) 2010 Elsevier B.V. All rights reserved.
Application-level checkpointing can decrease the overhead of fault tolerance by minimizing the amount of checkpoint data. However this technique requires the programmer to manually choose the critical data that should...
详细信息
ISBN:
(纸本)9781424431748
Application-level checkpointing can decrease the overhead of fault tolerance by minimizing the amount of checkpoint data. However this technique requires the programmer to manually choose the critical data that should be saved. In this paper, we firstly propose a live-variable analysis method for MPI programs. Then, we provide an optimization method of data saving for application-level checkpointing based on the analysis method. Based on the theoretical foundation, we implement a source-to-source pre-compiler (ALEC) to automate application-level checkpointing. Finally, we evaluate the performance we of five FORTRAN/MPI programs which are transformed and integrated checkpointing features by ALEC on a 512-CPU cluster system. The experimental results show that i)the application-level checkpointing based on live-variable analysis for MPI programs can efficiently reduce the amount of checkpoint data, thereby decrease the overhead of checkpoint and restart;ii)ALEC is capable of automating application-level check-pointing correctly and effectively.
Petri nets are commonly used for modeling of computer, communication, logistics, and manufacturing systems. Powerful grid and cluster computers allow efficient distributed simulation of such models to speed up executi...
详细信息
ISBN:
(纸本)0769524621
Petri nets are commonly used for modeling of computer, communication, logistics, and manufacturing systems. Powerful grid and cluster computers allow efficient distributed simulation of such models to speed up execution. Optimistic parallel simulation techniques have been developed for different Petri net classes. Different notions of time have been proposed to detect occurring causality violations. However global guards in timed colored Petri nets restrict state transitions based on global conditions. Furthermore those nets may contain vanishing states and transition priorities. Those properties require an extended logical time scheme and an efficient monitoring mechanism to capture global conditions over time. The paper presents an appropriate logical time scheme that guarantees correct ordering of global states in a distributed timed colored Petri net simulation. A new mirrored state mechanism is proposed to verify global conditions without limiting the optimistic simulation behavior Some Petri net examples are used to present performance values on a cluster computer.
Bioinformatics needs high-throughput computing and huge data Storage to understand datasets such as ones produced by complete genome projects. But these data are linked to patients, and used in scientific or industria...
详细信息
ISBN:
(纸本)0769525679
Bioinformatics needs high-throughput computing and huge data Storage to understand datasets such as ones produced by complete genome projects. But these data are linked to patients, and used in scientific or industrial processes such as drug design and gene function identification. These use cases need to have a certain level of confidentiality and integrity to preserve the patient privacy or the patent secret. Obviously important in a local computing context such as supercomputer or cluster, these requirements are exarcebated in the context of a grid such EGEE, where the computing and storage resources are distributed across a worldwide platform. We have developed the EncFile encrypted files management system, deployed on the production platform of the EGEE project, and associated to encrypted representative biological resources. Thus we provided grid users with a user-friendly component that doesn't require any user privileges. Moreover, our EncFile system can be also deployed on other distributed systems as it is not linked to the EGEE grid components.
This paper puts forwards a fast and collaborative search technology for massive and distributed spatial information based on grid dispatch and grid services, and the spatial search engine based on grid is designed. Th...
详细信息
ISBN:
(纸本)9781424473021
This paper puts forwards a fast and collaborative search technology for massive and distributed spatial information based on grid dispatch and grid services, and the spatial search engine based on grid is designed. The service of grid dispatch for massive and distributed spatial information and grid map services are implemented by using gird development tool ProActive. The efficiency of massive spatial information search is promoted by parallel and collaborative search. The instantaneous and dynamic spatial information acquiring for 300 Chinese cities is also achieved. The engine provides a new technology for massive spatial information service.
Communications and the scheduling of tasks are the two most important issues of parallel programming on clusters. Over time, various parallelcomputing programming models such as remote threads, transparent process mi...
详细信息
ISBN:
(纸本)1932415602
Communications and the scheduling of tasks are the two most important issues of parallel programming on clusters. Over time, various parallelcomputing programming models such as remote threads, transparent process migration, message passing, distributed shared memory and optimizing parallel compilers have emerged, assisting the programmer to develop applications which can work seamlessly in such environments. However their acceptance has varied by virtue of the fact that each has introduced new problems with reference to portability, scalability and usability. Sometimes these problems completely violate the underlying notion of such computing. This paper will examine the evolution of cluster computing and identify where deficiencies lie in current programming models, providing a justification for simple languages while providing the reader with an understanding of the Pshell programming environment we are actively developing. Pshell is the 'glue' for producing high-performance parallel applications that can work securely and efficiently in heterogeneous environments. In addition to this, the paper illustrates how such a language can be used to ease the process of writing parallel applications and to overcome some of the limitations inherent in traditional programming models without sacrificing performance.
The computinggrid infrastructure could benefit of techniques that can improve the overall throughput of the system. It is possible that job submission will include different ontology in resource requests due to the g...
详细信息
As HPC capability and software adaptability continues to expand, the interest to perform complex system-wide simulations involving multiple interacting components grows. In this paper, we present a novel integrative s...
详细信息
ISBN:
(纸本)9781479983346
As HPC capability and software adaptability continues to expand, the interest to perform complex system-wide simulations involving multiple interacting components grows. In this paper, we present a novel integrative software platform - the Distributive Interoperable Executive Library (DIEL) - to facilitate the collaboration, exploration, and execution of multiphysics modeling projects suited for a diversified research community on emergent large-scale parallelcomputing platforms. It does so by providing a managing executive, a layer of numerical libraries, a number of commonly used physics modules, and two set of native communication protocols. DIEL allows users to plug in their individual modules, prescribe the interactions between those modules, and schedule communications between them. The DIEL framework is designed to be applicable for preliminary concept design, sensitivity prototyping, and productive simulation of a complex system.
Load balancing is a key concern when developing parallel and distributedcomputing applications. The emergence of computational grids extends this problem, where issues of cross-domain and large-scale scheduling must ...
详细信息
暂无评论