以NaradaBrokering為基礎之運算能力分享環境

Abstract

隨著電腦的運算能力、儲存空間與網路科技持續的突破，不論在學界或在企業中，有越來越多的大型、複雜與資源密集的各種應用持續的開發出來。事實上，運算資源漸漸的追趕不上這些大型應用對他的需求，網路科技的成長使的分散在各處的資源得以聚集，而分散式運算也日益受到重視。在企業中，必須以最低的成本盡可能運用閒置的電腦資源以完成龐大的運算工作，網格運算(Grid Computing)已經成為企業內運算(Enterprise Computing)與企業間運算(Business-to-Business Computing)的關鍵科技。 Tsai et al. 2006所提出的Web Services Based Computing Power Service (WS-CPS) 系統結合網路服務與Oracle BPEL架構出企業內分散式運算環境，但是WS-CPS系統仍存在一些必須解決的問題。因此本文提出以NaradaBrokering為基礎的分散式運算環境 (NB-CPS)，適用於處理可切割並行運作的需要高度運算能力之工作，解決現有WS-CPS系統之彈性、容錯、工作分派效率與高檔案傳輸環境產生的不穩定性等問題。

As computation, storage, and communication technologies steadily improved, application developers and users are planning increasingly large runs that will operate on and generate petabytes of data. Although microprocessors are reaching impressive speeds, in the long run they are falling behind storage. Fortunately, improvements in wide-area networking make it possible to aggregate distributed resources in various collaborating institutions and to form what have come to be known as Computational Grids. Recently, Grid computing has been identified as a critical technology by industry for enterprise computing and business-to-business computing. Web Services Based Computing Power Service (WS-CPS) is a distributed computing system built for small and middle-sized enterprise. It combines Web Services and Oracle BPEL Services for providing a visualized developing environment and workflow management. However, there are unsolved issues in WS-CPS. Therefore, in this article, we proposed a NaradaBrokering Based Computing Power Service (NB-CPS), which is not only suitable for executing computing works which are able to run in batch, but also be able to solve current issues in WS-CPS such as system resilience, fault tolerance, efficiency of job scheduling and the instability in congested network environment.