A novel checkpoint mechanism based on job progress description for computational grid

Authors:
Chunjiang Lia;Xuejun Yang;Nong Xiao
Affiliations:
School of Computer, National University of Defense Technology, Changsha, China;School of Computer, National University of Defense Technology, Changsha, China;School of Computer, National University of Defense Technology, Changsha, China
Venue:
ISPA'04 Proceedings of the Second international conference on Parallel and Distributed Processing and Applications
Year:
2004

Citing 7
Cited 0

The grid: blueprint for a new computing infrastructure

The grid: blueprint for a new computing infrastructure
A grid-enabled MPI: message passing in heterogeneous distributed computing systems

SC '98 Proceedings of the 1998 ACM/IEEE conference on Supercomputing
Distributed Algorithms

Distributed Algorithms
Scalable Parallel Computing: Technology,Architecture,Programming

Scalable Parallel Computing: Technology,Architecture,Programming
MPICH-V: toward a scalable fault tolerant MPI for volatile nodes

Proceedings of the 2002 ACM/IEEE conference on Supercomputing
MPICH-G2: a Grid-enabled implementation of the Message Passing Interface

Journal of Parallel and Distributed Computing - Special issue on computational grids
Libckpt: Transparent Checkpointing under Unix

Libckpt: Transparent Checkpointing under Unix

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we argue that application-level uncoordinated checkpointing with user-defined checkpoint data is the favorable in grid environment where heterogeneity is essentially popular. We present a novel application-level uncoordinated checkpoint protocol based on Job Progress Description (JPD) which is composed by a Job Progress Record Object and a group of Job Progress State Objects, these two kinds of objects act as checkpoint data for the job and the methods of them can be used as checkpoint APIs. By extending this protocol with sender-based message logging, it can be used by the message passing applications in computational grid. Emulation with a kind of master-worker message-passing applications shows that using this checkpointing protocol can dramatically reduce the wall-time of the application when failure occurs.