Network workstations are widely accepted and applied to high-performance computing for their low-cost, efficient computing environment and fast and reliable services. Network workstations can be divided into dedicated networks and non-dedicated networks according to their user types. Dedicated workstations have the characteristics of relatively simple structure and management, easy expansion, etc. Non-dedicated workstations have the characteristics of making full use of computer resources, but the management is more complicated. This paper proposes a performance prediction mathematical model for non-dedicated network computing, which includes the utilization of a single workstation system and the distributed services of a homogeneous cluster system. Through this calculation model, the completion time of parallel tasks can be predicted from aspects such as machine utilization, continuous work service rate, and parallel task allocation. It provides a valuable reference for task scheduling in a non-dedicated network environment. Keywords: performance forecast model; cluster computing; non-dedicated network system; workload distribution abstract: The low cost and wide availability of networks of workstations have made them an attractive solution for high performance computing. According to the ownership, NOW can be divided into dedicated networks of workstations and non-dedicated networks of workstations. The structure and management are easier in dedicated NOW and is prone to expand. The non-dedicated NOW can make great use of computing resource, but theadministration is not easy. A performance forecast model of non-dedicated NOW computing is advanced in this paper. There are two parts―one workstation system\'s utilize and the distributed service of homogeneous cluster. The computing model result in the separation of the influence of machine utilization, sequential job service rate, and parallel task allocation to the parallel task completion time. It is a useful reference for non-dedicated networkenvironment.Key word: performance forecast model, cluster system, non-dedicated systems, workload distribution.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore