Efficient atomic broadcast using deterministic merge.pdfVIP

Efficient atomic broadcast using deterministic merge.pdf

  1. 1、有哪些信誉好的足球投注网站(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。。
  2. 2、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载
  3. 3、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
  4. 4、该文档为VIP文档,如果想要下载,成为VIP会员后,下载免费。
  5. 5、成为VIP后,下载本文档将扣除1次下载权益。下载后,不支持退款、换文档。如有疑问请联系我们
  6. 6、成为VIP后,您将拥有八大权益,权益包括:VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
  7. 7、VIP文档为合作方或网友上传,每下载1次, 网站将根据用户上传文档的质量评分、类型等,对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档
查看更多
Efficient atomic broadcast using deterministic merge

Efficient Atomic Broadcast Using Deterministic Merge Marcos Kawazoe Aguilera LIX Laboratoire d’Informatique E?cole Polytechnique 91128 Palaiseau Cedex, France aguilera@lix.polytechnique.fr Robert E. Strom IBM T.J. Watson Research Center Yorktown Heights, NY 10598 USA strom@ ABSTRACTWe present an approach for merging message streams fromproducers distributed over a network, using a determinis-tic algorithm that is independent of any nondeterminismof the system, such as the amount of time the messagesare delayed by the network, or their arrival order. Thus, ifthis algorithm is replicated at multiple \mergers, then eachmerger will merge the message streams in exactly the sameway. The technique is therefore a solution to atomic broad-cast and global atomic multicast [12]. We assume that eachproducer has access to (approximately) synchronized clocksand can estimate the expected message rates of all produc-ers. We propose an algorithm, called the Bias Algorithm.To measure the performance of the Bias Algorithm, we as-sume that messages are generated by memoryless processesoperating at known message rates, and we measure the ex-pected total merge delay at a given time L. For the case oftwo producer processes, we give optimal algorithms in thismetric, and show that the optimal algorithm converges toour Bias algorithm when L ! 1. We recon rm our opti-mality result using Dynamic Programming theory, and weuse simulations to validate the robustness of this optimalityresult under more realistic conditions.Our motivating application is middleware for a widely-dis-tributed publish-subscribe system, where subscribers requireuniform reliable ordered delivery. To allow recovery, mes-sages are logged to stable storage at \logger nodes, whichserve as the sources for the message streams to be merged.We discuss the practical bene ts of our approach in suchenvironments. 1. INTRODUCTION 1.1 Background and motivationConsider the system shown in Figure 1, derived from Gryphon,an experimental

文档评论(0)

l215322 + 关注
实名认证
文档贡献者

该用户很懒,什么也没介绍

1亿VIP精品文档

相关文档