{"title":"Scality's experience with a geo-distributed file system","authors":"M. Segura, Vianney Rancurel, Vinh Tao, M. Shapiro","doi":"10.1145/2678508.2678524","DOIUrl":null,"url":null,"abstract":"This paper presents our experience in building a large scale geo-distributed file system for a leading telecommunications service provider in France; the system, which spans two data centers in Paris and Lyon, is capable of storing 4.02PB of data. We use a structured P2P network protocol based on consistent hashing for scalability and load balancing. We combine different setups that ensure availability and consistency. We report the measurements, and issues encountered during one year of the production system; measurements show that our approach scales with real-world workloads.","PeriodicalId":412543,"journal":{"name":"Proceedings of the Posters and Demos Session of the 15th International Middleware Conference","volume":"39 9-10","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2014-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the Posters and Demos Session of the 15th International Middleware Conference","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/2678508.2678524","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 3
Abstract
This paper presents our experience in building a large scale geo-distributed file system for a leading telecommunications service provider in France; the system, which spans two data centers in Paris and Lyon, is capable of storing 4.02PB of data. We use a structured P2P network protocol based on consistent hashing for scalability and load balancing. We combine different setups that ensure availability and consistency. We report the measurements, and issues encountered during one year of the production system; measurements show that our approach scales with real-world workloads.