Tuple-Space分散式計算的例外處理不同於一般應用程式,其最大的不同在於系統間是透過訊息進行溝通,且難以得知整個系統的狀態,造成例外傳播與偵測上的困難;這使得現有的強健度等級提升策略並無法直接套用到提升Tuple-Space分散式系統強健度等級上。 在本論文中,首先分析Tuple-Space分散式計算中的主要例外情形,接著針對這些例外情形提出強健度等級提升的策略,並提供例外情形與策略的關聯表格,以引導開發人員進行分析與決定例外處理方式。最後透過實際案例,展示出如何根據需求選擇適當的策略,以做為維護或開發人員在提升系統強健度時的參考。 藉由本論文提出的方法,我們分別由系統架構面與實務面的角度,來協助開發人員進行Tuple-Space分散式計算例外處理的分析、設計,以期有效地增進軟體系統的強健度。
Tuple-space distributed computing is different from general application exception handling. The main differences are in the use of message exchange mechanism for the communication between each sub-system and the difficulty to obtain the global state in the system. These two differences introduce some difficulties in exception propagation and detection, which prevent the existing robustness improvement strategies from being applied in Tuple-space distributed systems. In this thesis, we will first analyze the main exception scenario in Tuple-space distributed computing, and then propose the corresponding strategies for improving robustness. Finally, we show the case studies are conducted to illustrate the effectiveness of the proposed strategies.