The Chubby lock service for looselycoupled distributed松散耦合的分布式锁服务胖乎乎的.pptVIP

下载本文档

6
0
约1.31万字
约 53页
2017-03-09 发布于上海
举报
版权申诉

The Chubby lock service for looselycoupled distributed松散耦合的分布式锁服务胖乎乎的.ppt

1、本文档共53页，可阅读全部内容。
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、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、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)。
5、该文档为VIP文档，如果想要下载，成为VIP会员后，下载免费。
6、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。
7、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。
8、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

The Chubby lock service for looselycoupled distributed松散耦合的分布式锁服务胖乎乎的

Design: Fail-overs In-memory state discarded sessions, handles, locks, etc. Lease timer “stops” Quick re-election client reconnect before leases expire Slow re-election clients disable cache, enter grace period allows sessions across fail-overs Design: Fail-overs Design: Fail-overs Steps of newly-elected master: Pick new epoch number Respond only to master location requests Build in-memory state for sessions / locks from DB Respond to KeepAlives Emit fail-over events to caches Wait for acknowledgements / session expire Allow all operations to proceed Design: Fail-overs Steps of newly-elected master (cont’d): Handle created pre-fail-over used master recreates in memory, honors call if closed, record that in memory Delete ephemeral files w/o open handles after an interval Fail-over code source of many bugs Design: Database First Chubby used replicated Berkeley DB with master lease added on Replication code was new did not want to take the risk Implemented own simple database distributed using consensus protocol Design: Backup Every few hours Snapshot of database to GFS server different building building damage, cyclic dependecies Disaster recovery Initialize new replica avoid load on in-service replicas Design: Mirroring Collection of files mirrored across cells Mostly for configuration files /ls/global/master mirrored to /ls/cell/slave global cell’s replicas spread around world Chubby’s own ACLs Files advertising presence / location pointers to Bigtable cells etc. Mechanisms for Scaling Clients individual processes (not machines) observed 90,000 clients for a single master Server machines identical to client ones Most effective scaling: reduce communication Regulate # of Chubby cells Increase lease time Caching Protocol-conversion servers Scaling: Proxies Proxies pass requests from clients to cell Can handle KeepAlives and reads Not writes, but they are 1% of workload KeepAlive traffic by far most dominant Disadvantages: additional RPC for writes / first time reads