A Domain-Specific Language for Describing Grid Applications

A Domain-Specific Language for Describing Grid Applications

Enis Afgan (University of Alabama at Birmingham, USA), Purushotham Bangalore (University of Alabama at Birmingham, USA) and Jeff Gray (University of Alabama at Birmingham, USA)
Copyright: © 2009 |Pages: 37
DOI: 10.4018/978-1-59904-699-0.ch013
OnDemand PDF Download:


Grid computing environments are dynamic and heterogeneous in nature. In order to realize applicationspecific Quality of Service agreements within a grid, specifications at the level of an application are required. This chapter introduces an XML-based schema language (called the Application Specification Language, ASL) and a corresponding modeling tool that can be used to describe applications in grid computing environments. Such application descriptions allow derivation of guided and autonomic service developments for installation and invocation routines throughout the grid. In order to promote the language and ease the application description process, a domain-specific tool is also introduced. Based on our experience, the ASL in combination with higher level models improves, simplifies and promotes the grid application deployment process while simultaneously minimizing tedious and error-prone tasks such as manual application description composition.
Chapter Preview


Grid computing (Foster, Kesselman, & Tuecke, 2001) has gained popularity as the emerging architecture for next-generation high performance distributed computing. Grid computing provides ubiquitous access to distributed high performance computing (HPC) resources that are shared between multiple organizations through virtualization and aggregation. This is realized through a layer of software (e.g., grid middleware), thus making grid applications extremely software-intensive systems where the value of software is equivalent to the value of the underlying infrastructure. Grid middleware provides a standard set of services for authentication, authorization, resource allocation and management, job scheduling, submission, monitoring, and data transfer and management (Berman, Hey, & Fox, 2003b). Software packages and tools based on open-source/open-standard approaches such as Globus Toolkit (Foster & Kesselman, 1997) have enabled the deployment of “production quality” computational grids. Several domain-specific grids are currently operational, for example, Grid Physics Network (GriPhyN) (GriPhyN, 2006), Network for Earthquake Engineering Simulation (NEESGrid) (NEES, 2006), International Virtual Data Grid Laboratory (IVDGL) (IVDGL, 2006), Open Science Grid (Grid, 2007), and Particle Physics Data Grid Collaboratory Pilot (Grid2003) (PPDG, 2006). Grid computing has offered researchers enormous computing and data storage capabilities by providing seamless access to geographically distributed resources through the creation of virtual organizations (VOs).

Despite the many benefits of grid computing, the grid itself does not provide a novel programming paradigm for developing new applications. Furthermore, no formal methodology exists for porting existing legacy applications to the grid. Most of the applications developed for the grid are based on traditional HPC or distributed computing principles. Typical HPC applications are developed using implicit parallel programming techniques (e.g., compiler-based automatic parallelization and directive-based parallelization) or explicit parallel programming techniques (e.g., threads and message-passing). After an HPC application is developed and tested on local resources, it is then deployed; that is, the entire application (source code, dataset, scripts) is transferred to a remote site, compiled on a remote host, and made available for execution.

Deploying an application on the grid requires additional steps that involve user intervention, great insight into the internal structure of the application, as well as familiarity with the various grid computing technologies and toolkits. In addition, the progressive steps of application execution and job submission may involve many additional steps required from the end-user, not necessarily found in a typical application. Due to this inherent complexity and difficulty using the grid, several approaches have attempted to simplify grid deployment and configuration by developing technologies such as Web portals (Gannon et al., 2003), workflow systems (Aalst & Hee, 2002), and component assembly (Armstrong et al., 1999). The ultimate goal of such efforts is to enable the adoption of grid technologies and applications to a wider group of end-users who are not familiar with programming languages and the lower level grid infrastructure. The potential impact for improving grid accessibility to such users is significant (e.g., applied science researchers, distributed organizations, and organizations with variable computational requirements).

Complete Chapter List

Search this Book: