globus數(shù)據(jù)管理服務(wù)-文庫吧資料
2024-10-20 15:50本頁面
【正文】 肖儂 GASS Architecture Cache GASS Server HTTP Server FTP Server % globusgasscache (c) Remote cache management GRAM (a) GASS file access API amp。網(wǎng)格計(jì)算－ Grid Computing 肖儂 Globus數(shù)據(jù)管理服務(wù) 網(wǎng)格計(jì)算－ Grid Computing 肖儂 數(shù)據(jù)管理服務(wù) ? 數(shù)據(jù)傳和訪問 ?GASS: Provides services mainly intended for use with GRAM (file staging, I/O redirection) ?GridFTP: Provides highperformance, reliable data transfer for modern WANs ? 數(shù)據(jù)復(fù)制和管理 ?復(fù)制目錄 : Provides a catalog service for keeping track of replicated datasets ?復(fù)制管理 : Provides services for creating and managing replicated datasets 網(wǎng)格計(jì)算－ Grid Computing 肖儂 Globus－ Gridftp 網(wǎng)格計(jì)算－ Grid Computing 肖儂 公共數(shù)據(jù)訪問協(xié)議目的 ? 已存在的分布數(shù)據(jù)存儲系統(tǒng) ?DPSS, HPSS: focus on highperformance access, utilize parallel data transfer, striping ?DFS: focus on highvolume usage, dataset replication, local caching ?SRB: connects heterogeneous data collections, uniform client interface, metadata queries ? 問題（ 協(xié)議不兼容、不公開，各自獨(dú)立的客戶端） ?不兼容的協(xié)議和特性 ? Each require custom client ? Partitions available data sets and storage devices ?每一個(gè)協(xié)議有所希望功能的一部分 網(wǎng)格計(jì)算－ Grid Computing 肖儂 需要公共安全有效的數(shù)據(jù)訪問協(xié)議 ? 公共可擴(kuò)展的傳輸協(xié)議 ?意味著可以互操作 ? 從存儲服務(wù)中分離低級的傳輸機(jī)制 ? 優(yōu)點(diǎn) : ?新的特殊的存儲系統(tǒng)可以自動地和已經(jīng)存在系統(tǒng)兼容 ?已經(jīng)存在系統(tǒng)具有豐富的數(shù)據(jù)傳輸功能 ? 和多個(gè)存儲系統(tǒng)接口 ?HPSS, DPSS, file systems ?Plan for SRB integration 網(wǎng)格計(jì)算－ Grid Computing 肖儂 Globus提出統(tǒng)一 GridFTP協(xié)議 ? 基于 FTP ?與大量已有工具兼容 ?已經(jīng)支持不少數(shù)據(jù)網(wǎng)格需要的特征，容易擴(kuò)展 ?普遍認(rèn)可、理解與支持 ? 已有的規(guī)范 ?RFC 949: File Transfer Protocol ?RFC 2228: FTP Security Extensions ?RFC 2389: Feature Negotiation for the File Transfer Protocol ? GridFTP包括什么？ ?協(xié)議 ?協(xié)議實(shí)現(xiàn)的一套工具 ? GridFTP FTP，是 FTP的超集， GridFTP不僅僅限于文件傳輸 網(wǎng)格計(jì)算－ Grid Computing 肖儂 GridFTP: 基本方法 ? 從最通用的子集開始 ?Standard FTP: get/put etc., 3rdparty transfer ? 實(shí)現(xiàn)標(biāo)準(zhǔn)化但不經(jīng)常使用的特性 ?GSS binding, extended directory listing, simple restart ? 多個(gè)方面的擴(kuò)展，但保持與已存在服務(wù)器互操作能力 ?Striped/parallel data channels, partial file, automatic amp。 manual TCP buffer setting, progress monitoring, extended restart 網(wǎng)格計(jì)算－ Grid Computing 肖儂 GridFTP的特點(diǎn) ? 互操作，可擴(kuò)展 ? 兩個(gè)分離 ?底層數(shù)據(jù)傳輸機(jī)制和數(shù)據(jù)存儲服務(wù)分開 ?將控制通道和數(shù)據(jù)通道分離 ? 繼承 FTP的通用性和廣泛性 ?FTP is defined by several IETF（ Inter Engineering Task Force） RFCs 網(wǎng)格計(jì)算－ Grid Computing 肖儂 GridFtp對 FTP的擴(kuò)展 ? 安全 GSI （ Publickeybased Grid Security Infrastructure ) or Kerberos support (via GSSAPI) ?GridFTP provides this capability by implementing the GSSAPI authentication mechanisms defined by RFC 2228, “FTP Security Extensions”. ? 第三方控制數(shù)據(jù)傳輸 ?a “thirdparty” user or application at one site to initiate, monitor and control a data transfer operation between two other “parties” ? 并行數(shù)據(jù)傳輸 ?using multiple TCP streams in parallel (even between the same source and destination) can improve aggregate bandwidth over using a single TCP stream. 網(wǎng)格計(jì)算－ Grid Computing 肖儂 GridFtp對 FTP的擴(kuò)展 ? 分片數(shù)據(jù)傳輸 ?use multiple TCP streams to transfer data that is partitioned among multiple servers. Striped transfers provide further bandwidth improvements over those achieved with parallel transfers. ? 部分文件傳輸 ?transferring portions of files rather than plete files. ? 自動協(xié)商設(shè)置 TCP buffer/window的大小 ?Using optimal settings for TCP buffer/window sizes can have a dramatic impact on data transfer performance. ? 數(shù)據(jù)可靠傳輸
教學(xué)課件相關(guān)推薦
鄂ICP備17016276號-1