網格計算是將分散在各地的異質資源，不管是不同的平台、軟硬體、電腦架構、電腦語言以及不同管理網域，都可以透過網路以及公開標準串聯成一虛擬計算資源，並用來解決一些需要強大計算能力的科學問題。然而，當網格環境擴充到一定程度時，要充分發揮它的能力以及潛力，在上面執行的程式勢必會利用到跨網域的資源，而不再只是一個叢集系統可以應付的了。本論文提出一個在網格計算環境上的跨站資源配置策? (MSRA) 給資源代理人 (Resource Broker) 在分配工作時，可以考慮如何分配適當的跨站資源給該工作，實驗結果顯示本策略明顯優於另外兩種的策略方法。除此之外，本論文提出一個以網域為基礎的網路資訊模型，該模型利用 NWS (Network Weather Service) 達到有效減少收集時的額外頻寬浪費且提供真實網路資訊給網格環境。再者，我們透過利用 Ganglia 以及 NWS 監控資源狀態以及網路相關資訊，使得資源代理人 (Resource Broker) 可提供一個可用資源的即時訊息，並扮演一個網格環境中互異系統的聯繫角色，達到統一存取的目的。 Grid computing encounters distributed heterogeneous resources, including different platforms, hardware/software, computer architecture, and computer languages, which are geographically distributed and governed by different Administrative Domains over a network using open standards to solve large-scale computational problems. As more Grids are deployed worldwide, the number of multi-institutional collaborations is rapidly growing. However, to realize Grid computing full potential, it is expected that Grid participants are able to use one another’s resources. This work presented a multi-site resource allocation (MSRA) strategy for Resource Broker to dispatch jobs to appropriate resources across two different Administrative Domains and the experimental result shown that MSRA exhibits a better performance than other strategies. In this work, we addressed information gathering and focused on providing a domain-based model for network information measurement using Network Weather Service (NWS) on Grid computing environments. We used the Ganglia and NWS tools to monitor resource status and network-related information, respectively. The proposed broker provided secure, updated information about available resources and served as a link to the diverse systems available in the Grid.
