Analyzing metadata performance in distributed file systems [Elektronische Ressource] / vorgelegt von Christoph Biardzki

Inaugural-DissertationzurErlangungderDoktorwur¨ dederNaturwissenschaftlich-MathematischenGesamtfakultat¨derRuprecht-Karls-Universitat¨ HeidelbergvorgelegtvonDiplom-InformatikerChristophBiardzkiausThornTagdermundlichen¨ Prufung:¨ 19.1.2009AnalyzingMetadataPerformanceinDistributedFileSystemsGutachter: Prof. Dr. ThomasLudwigAbstractDistributedfilesystemsareimportantbuildingblocksinmoderncomputingenvironments.The challenge of increasing I/O bandwidth to files has been largely resolved by the useof parallel file systems and sufficient hardware. However, determining the best means bywhich to manage large amounts of metadata, which contains information about files anddirectoriesstoredinadistributedfilesystem,hasprovedamoredifficultchallenge.Theobjectiveofthisthesisistoanalyzetheroleofmetadataandpresentpastandcurrentimplementations and access semantics. Understanding the development of the current filesysteminterfacesandfunctionalityisakeytounderstandingtheirperformancelimitations.Basedonthisanalysis,adistributedmetadatabenchmarktermedDMetabenchispresented.DMetabench significantly improves on existing benchmarks and allows stress on meta-data operations in a distributed file system in a parallelized manner. Both intra-node andinter-node parallelity, current trends in computer architecture, can be explicitly tested withDMetabench. Thisisduetothefactthatadistributedfilesystemcanhavedifferentseman-ticsinsideaclientnoderatherthansemanticsbetweenmultiplenodes.