The Research And Realization Of The Distributed Object Name System

With the rapid development of Internet services and mobile Internet, the enterprisesystem has accumulated vast amounts of data. Now, many large Internet companies are facingwith a huge technical problem-how to store and process the massive data. In order tofacilitate the management of these massive data, we divide these massive data into severalparts, each part call a Domain. And each data in the domain call an Object. For each object, ithas a unique identifier, the unique identifier is string that make of0and1and its length is512bits. In order to remember the object easily, a distributed object name is given to each object.But when you actually need to access an object, the object cannot be accessed through thename, but can only be accessed through the unique identifier. Therefore, we propose aDistributed Object Name System, which is a name service system likes the Domain NameSystem. It’s mainly task is transformed a distributed object name into the unique identifier ofthe object. The Distributed Object Name System can help us realize the unified managementof the object of various distributed systems. Therefore, there is no doubt that this research ismeaningful and useful.This paper studies the key technologies of Distributed Object Name System and therealization of Distributed Object Name System prototype.Firstly, we have a reach on the basic theory and the key technologies, such as the uniqueidentification technologies of the commodity logistics, the key technologies of the DomainName system and the key technologies and theory of data caching. And then, we design thedistributed object name and the structure of the object unique identifier.Secondly, we analysis the query cases of the Distribute Object Name System anddesigned the query algorithms. To speed up the query response time, we have introduced datacaching technology. We studies and compare a variety of cache replacement algorithms, suchas LRU replacement algorithm, FIFO replacement algorithm, the MRU replacementalgorithm, LFU replacement algorithm, LRFU replacement algorithm and LRU-Kreplacement algorithm. What’s more, we also study the data synchronization method, such asTTL method, broadcast writing method, each recognized method and directory table method.And then combine with the Distributed Object Name System characteristics; we select the LRU cache replacement algorithms and TTL cache synchronization algorithm to ouralgorithms. Besides we design our cache model.Finally, we design and realize the prototype of DONS. What’s more, we design suitableexperimental methods, testing the concurrent access to stress and query response time of theDistributed Object Name System. And then give a conclusion that whether meet the needs ofpractical application.