Font Size: a A A

The Design And Implementation Of Distributed And Parallel Computing Model

Posted on:2007-05-12Degree:MasterType:Thesis
Country:ChinaCandidate:Q K JinFull Text:PDF
GTID:2178360182460895Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Because the network bandwidth and the performance of computer improve sustainedly, there are more and more networked personal computers in lab, enterprise and government. There are great computing capabilities hidden in these computers. If these organizations can make the most of these computers in their computational capabilities, they will not only avoid the waste, but also improve the computing power as a whole. Hence enterprises can reduce their IT cost and improve their competitive capacity. Distributed and parallel computing has become one of best methods to utilize these existing idle computing resources, while the large concurrent processor is expensive and not userfriendly, so recently network parallel computing that is cheaper has become the focus studied by more and more researcher.In order to make use of idle computing resource in local network thoroughly, this paper firstly analyze the mechanism of SUN RPC; then design and implement DPCM-Distributed and Parallel Computing Model which uses much technology of computer, such as TCP/IP, Multi-Thread, Object Serialization and Load balance and so on. In the processing of designing this model, I divide the function of this model into some modules, then implements these modules respectively; Lastly, this paper make some test scheme to test the performance of this model.This model has three properties: modularity which simplifies the upgrading and maintaining of this model, and provides the reusable modules for other similar systems; Expansibility which makes this model work together well with other works studied by lab; Userfriendliness which liberate this model users from the burden of learning many other related knowledge. Viewing the results of the performance test of this model, the efficiency of this model is high. After adding the function of load balance to this model, this model can find the idlest node in local network to compute. This model is the base of the other distributed study in the lab.
Keywords/Search Tags:Distributed and Parallel Computing Model, Parameter marshalling, Remote Procedure Call, Load balance
PDF Full Text Request
Related items