{"title":"用于重叠广播的具有多启动器的因果相位排序协议","authors":"L. Amanton, M. Naimi","doi":"10.1109/LCN.1999.802053","DOIUrl":null,"url":null,"abstract":"Distributed applications have often to determine consistent cuts (global states) in order to ensure virtual synchrony between independent tasks. Moreover most of the sequential programs can be cut out in more or less independent tasks; we group these tasks into phases to allow an asynchronous computation. The processes exchange messages without a particular ordering inside each phase. But messages of two successive phases must be ordered according to the causal relation. The paper presents the concept of causal-phase and an algorithm which guarantees that the change of phase of a distributed system is safely performed, i.e., that the causal ordering of the messages events is being respected. In contrast to previous approaches, our protocol is able to handle the multi-initiator case were the change of phase is simultaneously requested by several subprocesses without requiring any kind of master process. This special scheduling establishes a synchronization between two distinct phases, that may overlap, while preserving the asynchrony of the communications, without blocking the processes.","PeriodicalId":265611,"journal":{"name":"Proceedings 24th Conference on Local Computer Networks. LCN'99","volume":"89 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1999-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":"{\"title\":\"A causal-phase ordering protocol with multi-initiators for overlapped broadcasts\",\"authors\":\"L. Amanton, M. Naimi\",\"doi\":\"10.1109/LCN.1999.802053\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Distributed applications have often to determine consistent cuts (global states) in order to ensure virtual synchrony between independent tasks. Moreover most of the sequential programs can be cut out in more or less independent tasks; we group these tasks into phases to allow an asynchronous computation. The processes exchange messages without a particular ordering inside each phase. But messages of two successive phases must be ordered according to the causal relation. The paper presents the concept of causal-phase and an algorithm which guarantees that the change of phase of a distributed system is safely performed, i.e., that the causal ordering of the messages events is being respected. In contrast to previous approaches, our protocol is able to handle the multi-initiator case were the change of phase is simultaneously requested by several subprocesses without requiring any kind of master process. This special scheduling establishes a synchronization between two distinct phases, that may overlap, while preserving the asynchrony of the communications, without blocking the processes.\",\"PeriodicalId\":265611,\"journal\":{\"name\":\"Proceedings 24th Conference on Local Computer Networks. LCN'99\",\"volume\":\"89 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"1999-10-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"2\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Proceedings 24th Conference on Local Computer Networks. LCN'99\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/LCN.1999.802053\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings 24th Conference on Local Computer Networks. LCN'99","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/LCN.1999.802053","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
A causal-phase ordering protocol with multi-initiators for overlapped broadcasts
Distributed applications have often to determine consistent cuts (global states) in order to ensure virtual synchrony between independent tasks. Moreover most of the sequential programs can be cut out in more or less independent tasks; we group these tasks into phases to allow an asynchronous computation. The processes exchange messages without a particular ordering inside each phase. But messages of two successive phases must be ordered according to the causal relation. The paper presents the concept of causal-phase and an algorithm which guarantees that the change of phase of a distributed system is safely performed, i.e., that the causal ordering of the messages events is being respected. In contrast to previous approaches, our protocol is able to handle the multi-initiator case were the change of phase is simultaneously requested by several subprocesses without requiring any kind of master process. This special scheduling establishes a synchronization between two distinct phases, that may overlap, while preserving the asynchrony of the communications, without blocking the processes.