دانلود مقاله ISI انگلیسی شماره 21878
ترجمه فارسی عنوان مقاله

سیستم جریان کار مجازی برای برنامه های علمی و همکاری های توزیع شده در شبکه

عنوان انگلیسی
Virtual workflow system for distributed collaborative scientific applications on Grids
کد مقاله سال انتشار تعداد صفحات مقاله انگلیسی
21878 2011 11 صفحه PDF
منبع

Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)

Journal : Computers & Electrical Engineering, Volume 37, Issue 3, May 2011, Pages 300–310

ترجمه کلمات کلیدی
- محاسبات شبکه - پیاده سازی - سیستم جریان کار
کلمات کلیدی انگلیسی
Grid computing, implementation,workflow system
پیش نمایش مقاله
پیش نمایش مقاله  سیستم جریان کار مجازی برای برنامه های علمی و همکاری های توزیع شده در شبکه

چکیده انگلیسی

Grid computing has become an effective computing technique in recent years. This paper develops a virtual workflow system to construct distributed collaborative applications for Grid users. The virtual workflow system consists three levels: abstract workflow system, translator and concrete workflow system. The research highlight of the implementation is that this workflow system is developed based on CORBA and Unicore Grid middleware. Furthermore, this implementation can support legacy application developed with Parco and C++ codes. This virtual workflow system can provide efficient GUI for users to organize distributed scientific collaborative applications and execute them on Grid resources. We present the design, implementation, and evaluation of this virtual workflow system in the paper.► In this paper, we propose the methodology of virtual workflow solution by mapping abstract workflow components to concrete application services (such as CORBA object). ► We design a virtual workflow system that includes an abstract workflow system, concrete workflow system, registry service and a GUI. ► We implement the virtual workflow system on real Grid computing environment with Unicore Grid middleware and PACO application services..

مقدمه انگلیسی

Grid computing [1] is one of the most influential aspects of computing techniques in recent years. Distinguished from conventional parallel and distributed computing, Grid computing mainly focuses on resource sharing among geographically distributed sites and the development of innovative, high performance oriented applications. Computational Grids can present Grid users with pervasive and inexpensive access to a wide variety of resources. Most of current Grid users are scientists and engineers from research fields other than computer science, e.g., bio-informatics, particle physics, and aerodynamics. They are not familiar with Grid techniques, such as resource selection, task preparation and submission. Thus, a friendly user layer which hides a Grid’s technical details for end-users is required. Furthermore, Grid infrastructures are required to support heterogeneous middleware and application environments, for example, legacy applications and systems. The collaborative applications are one of most important applications from scientific and engineering fields. Many tasks running in Grid environments will be executed on multiple sites by several collaborators [2]. This paper is devoted to the new modeling and executing environments of distributed collaborative applications. A virtual workflow system is designed and implemented for the purpose. Users can construct the collaborative application in the virtual workflow system without detailed Grid knowledge. To justify our work, we evaluate a use case from the aerodynamics research field. The contribution of this paper includes: • We develop flexible and collaborative workflow models that support multiple workflow implementations and Grid middleware. • We develop a virtual workflow system that can decouple the workflow collaboration processes, definition and execution. • We demonstrate our work with a CORBA based legacy application on Unicore Grids. The rest of this paper is organized as follows: Section 2 introduces the background and related work of Grid computing, Grid workflow and distributed collaborative applications. Section 3 defines research issues for developing collaborative applications on Grids and our proposed solutions. Sections 4 and 5 presents our collaborative workflow model and the implementation of virtual workflow system. A use scenario is selected and examined in Section 6 to justify the system and Section 7 concludes the paper.

نتیجه گیری انگلیسی

Grid computing has become an effective computing paradigm for distributed high performance computing applications. Although it has been well developed, in our Grid computing project, we still meet some research challenges, for example, develop Grid middleware to support legacy parallel CORBA applications. Furthermore, our collaborative applications require development/deployment environments, which can high underlying Grid hardware/software complexities and provide an easy interface for end users. Aimed to solve above research issues, This paper is devoted to the research on developing a virtual workflow system on computational Grids for distributed collaborative applications. To support legacy parallel CORBA application, we have developed Unicore plug-ins that enables Grid middleware, Unicore in our work, to manage parallel CORBA codes. To provide an easy interface for developing collaborative application on Grid, this paper presents a virtual workflow system, which is flexible to be adopted in computational Grids with various middleware and execution environments. We decouple the workflow coordination, workflow specification and workflow execution by introducing multiple levels in the virtual workflow system: GUI and abstract workflow system, translator, registry and concrete workflow system. With the abstract workflow and GUI, users can construct a new workflow application without detailed information of underlying Grid resources. The concrete workflow system is developed based on parallel CORBA system (PaCO) so it can support legacy codes with CORBA such as Ag2D. The translator and registry service are developed for translating abstract workflow model to the concrete workflow model. We test our solutions and implementations on the real test bed that contains multiple clusters. A real parallel CORBA test case is used as the test case. We tested the task turn around time for workflow execution and the speedup of parallel workflow execution. The task execution and its performance justified the our design and implementation.