The Implementation of a Distributed Object Manager

• Main Authors: Shyh-Dong Kuo, 郭世棟
• Other Authors: Prof. L.M. Tseng
• Format: Others
• Language: zh-TW
• Published: 1993
• Online Access: http://ndltd.ncl.edu.tw/handle/15914264655841371706

碩士 === 國立中央大學 === 資訊及電子工程研究所 === 81 === Recently, personal workstation and local area network come to life of people. This paper attempts to combine this type of computing power to let people to construct a distributed application by use of an object based system. This system which is called distributed object manager can let people create an object, destroy an object, migrate an object, etc. The RPC ( Remote Procedure Call) is used as a message passing mechanism between any two objects and we establish the system using the client-server model. In this paper, we first introduce the life cycle of an object to discuss the state transition of an object, then we discuss the activities of object management in detail, including long, short and medium life cycle model, object naming, object creation, object destruction, object invocation, object migration and object locating. There are many problems in this type of system we must solve. Because it is important to keep track of objects in this system, we use an object name server to serve as this purpose. The object name server keeps all the informations of an object. Checkpointing lets the object survive the disaster when the machine crashes. When an object is migrated, there may be an invocation failure. Besides the object name server is used to keep track of the location of an object, forwarding address and the object locating algorithm are used to locate the invoked object because it may have been migrated. At last, we show our experiences of constructing this object manager. We will discuss and compare the pros and cons of its methods.