Caching and scheduling for broadcast disk systems.pdf

Caching and scheduling for broadcast disk systems.pdf

  1. 1、本文档共26页,可阅读全部内容。
  2. 2、原创力文档(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。
  3. 3、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
  4. 4、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)。
查看更多
Caching and scheduling for broadcast disk systems

Caching and Scheduling for Broadcast Disk SystemsVincenzo Liberatore1Institute for Advanced Computer Studies 1UMIACS, A. V. Williams Building, University of Maryland, College Park, MD 20742. E-mail:vliberatore@. URL: /users/liberato/. AbstractUnicast connections lead to performance and scalability problems when a large client population attemptsto access the same data. Broadcast push and broadcast disk technology address the problem by broadcastingdata items from a server to a large number of clients. Broadcast disk performance depends mainly on cachingstrategies at the client site and on how the broadcast is scheduled at the server site. An on-line broadcastdisk paging strategy makes caching decisions without knowing access probabilities. In this paper, we subjecton-line paging algorithms to extensive empirical investigation. The Gray algorithm [25] always outperformedother on-line strategies on both synthetic and Web traces. Moreover, caching limited the skewness neededfrom a broadcast schedule, and led to favor ecient caching algorithms over re ned scheduling strategieswhen the cache was not small. Prior to this paper, no work had empirically investigated on-line pagingalgorithms and their relation with server scheduling. 1 IntroductionThe demand of network data services has been growing exponentially during the past few years. Moreand more often, increased workloads cannot be satis ed by current technology. In particular, when clientsrequested several million connections to a hot Web site during peak periods (e.g. Deep Blue chess match,Olympic games), servers were overmatched by the heavy workload. The problem was that point-to-point(unicast) connections satis ed each request individually, and server performance did not scale with thenumber of requests. In general, the point-to-point paradigm poses a scalability problem that is exacerbatedby the exponential and sustained growth of data service demand. In this scenario, broadcast push promises toaddress the issu

文档评论(0)

l215322 + 关注
实名认证
内容提供者

该用户很懒,什么也没介绍

1亿VIP精品文档

相关文档