Volume 2, Issue 6 (Dec 2004)
DISTRIBUTED COMPUTING ISSUE
 Home :  Feedback
   Browse   
Search  in
 

MCCF: A Distributed Grid Job Workflow Execution Framework
Principal Investigator:  Cai Wentong

E-mail: aswtcai@ntu.edu.sg
With the explosion of scientific data, distributed scientific applications present great challenges to the existing job workflow execution models over the Grid. Based on the idea of having executable codes as part of Grid resources, a Mobile Code Collaboration Framework (MCCF) utilizing light-weight mobile agents and dynamic services for distributed job workflow execution is investigated in this project. The Light-weight Mobile Agent (LMA) technique is used to adapt job workflow execution to the dynamic characteristics of the Grid and to reduce the security risks. In addition, a dynamic service mechanism is also introduced to facilitate the multidisciplinary scientific cooperation and application integration over the Grid. A job workflow specification includes static specification and dynamic specification. The static specification specifies the sub-jobs, data dependencies among sub-jobs, and the user policies for resource scheduling. It is generated from the user input. The dynamic specification is generated during job workflow execution, and includes locations of computational resources for sub-job execution, locations of the required data set and code. The objective of MCCF is to map the static job workflow specification to the dynamic Grid resources on the fly for distributed job workflow execution, as shown in the figure below. The LMA in MCCF is defined using an Agent Core (AC), which is an XML file specifying the functional descriptions of sub-jobs. This is different from the light-weight mobile agents generally used in existing systems. An AC will migrate from resource to resource. Executable code is provided as dynamic services and is downloaded to the resource and executed according to demand. Common non-functional code is provided as a group of agents that work cooperatively for sub-job execution and resource scheduling. These agents include the schedule agent, task agent, partner agent, and coordination agent. They are constructed on each host when the AC arrives and are destroyed when the AC is migrated or disposed. As a proof-of-concept, a prototype of MCCF is implemented using JADE (http://jade.tilab.com/) and Globus 2.2 (http://www.globus.org). “A Mobile Code Collaboration Framework (MCCF) utilizing light-weight mobile agents and dynamic services for distributed job workflow execution is investigated in this project.” Principal Investigator: Cai Wentong E-mail: aswtcai@ntu.edu.sg Job Workflow Execution over the Grid
PRINT ARTICLE | EMAIL ARTICLE TO FRIEND
Dec2004

  • NANYANG CAMPUS GRID Accounting System
  • DESIGN OF A GRID-ENABLED LOW-COST SUPERCOMPUTER WITH MASSIVELY PARALLEL ACCELERATORS
  • MCCF: A Distributed Grid Job Workflow Execution Framework
  • AN INTEGRATED AND ADAPTIVE DECISION SUPPORT FRAMEWORK FOR HIGH-TECH MANUFACTURING AND SERVICE NETWORKS
  • DELTA: A VIRTUAL TRAINING ENVIRONMENT FOR MOUT
  • TIME-CONSTRAINED DATA BROADCAST SCHEDULING
  • COLLABORATIVE OFFICE DOCUMENT EDITING AND REVISION CONTROL
  • DS-Grid: Large Scale Distributed Simulation on the Grid
    •   Best viewed with Microsoft IE 6.0 and Netscape 6.0 and above
    Copyright© NTU School of Computer Engineering    | NTU Home | Manage My Subscription