Font Size: a A A

Research On Message Log Recovery Algorithm Based On Message Reordering And Message Number Check

Posted on:2014-01-26Degree:MasterType:Thesis
Country:ChinaCandidate:X Q WangFull Text:PDF
GTID:2268330425962218Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the rapid growth of computer technology, more and more people will use the distributed systems to replace the original systems. But the distributed systems are still not stable enough.In addition, the environmental factors and artifical factors result in system breakdown easily. The system breakdown is bound to affect people’s normal life. So we should resolve the problem of how to maintain the distributed system’s normal operation. To achieve the goal, we should know the node which can’t make the system breakdown when the node occurs error. And we should know the reason of system breakdown. The most important point is that we should know how to make system resume. The fault tolerant technology can give us a good solution. So in our work, we main work is around this topic.This research is sponsored by Natural Science Foundation of Shandong Province named "the research and implementation of fault tolerant technology on the distributed system based the backward resumption". This article firstly introduces the importancee of check point technology and the development status of this area. Then this article analyzes the possible problem of distributed system. According to the problem, we proposed that the system should be global consistant. And we should take technolgy methods to prevent the process block of check point setting and roll-back resumption. Through these methods, we can make the number of message in the reasonable range. The article descripes three kind of checking-point protocol and three kind of message log protocol. Because of the loss of message, the disorder of message, midway message and repeating message, the system can’t make the check point consistent. So the system can’t get the correct computation result. Meanwhile there is asynchrony between the preservation of optimism log and process communication. When the process ocurrs fault, the system can’t receive the order message and the disorder message will make the system not running properly. To resolve this problem, this article proposed the message log recovery algorithm which is based on message reordering and message number check.
Keywords/Search Tags:distributed systems, fault-tolerance, checkpoints, message logging, rollback recovery
PDF Full Text Request
Related items