一個在 CORBA 上的可容錯物件服務

Abstract

隨著分散式系統及物件導向技術的發展，以物件為基礎的分散式物件 處理已漸受重視。因此， OMG (Obect Management Group) 致力於分散式 物件交換標準的制定，而制定了CORBA (Common Object Request Broker Architecture)。 目前在 CORBA 的環 境上制定了愈來愈多的物件服務標準 (COSS -- Common Object Service Specification)，但是現今並沒有有關可容錯物件服務的標準制定；由於 在一個大型分散式應用程式中的物件必須要考慮到可能會發生的錯誤，如 傳輸的延遲，網路連接的中斷等等，所以已經有不少在 CORBA 上增加可 容錯能力的研究，如 ELECTRA，Orbix+ ISIS 等等。 在這篇論文中，提出了一個以重覆物件的方式來實現的可容錯物件服務設 計，當主要物件被調用時，它會同時調用一個次級物件，並且將自已的狀 態和收到的請求記錄在次級物件中，如果主要物件發生錯誤，次級物件會 自動接手，成為新的主要物件。 在實作上，我們依照 COSS 的方式定義了一個程式庫，程式員可以利用所提供的介面製作出具 有錯能力的的應用程式。此程式庫已經實作在一個符合 CORBA 規格的實 際產品 Orbix 上。 With the development of distributed systems and object- oriented technology, object-based distributed processing has been highly valued. Therefore, OMG (Object Management Group) is devoting itself to the standardization of distri-buted object exchange, and CORBA (Common Object Request Broker Architecture) is exactly its first standard for distributed object exchange. There are more and more COSS (Common Object Service Specification) on CORBA,but no common specification about fault- tolerance exists now. Since classes of large-scale distributed applications will have to deal with unpredictable communication delays, with partial failures, and with networks partitioning, there have been many reseraches for fault-tolerance on CORBA environment such as ELECTRA, Orbix+ISIS, etc. We propose a "warm stand-by" replication approach in this thesis. When an object (the primary object) is invoked, it will invoke a secondary object, andthe primary object will log the messages and checkpoint the state to the se- condary object periodically. If the primary object fails, the secondary can take over automatically. We provide a library according to the style of COSS.With such an interface, programmers can write programs with fault-tolerance a-bility. The whole model has been implemented on Orbix, which is a full imple-mentation of CORBA specification.