DOCSLIB.ORG

Dateisysteme

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Dateisysteme Dateisysteme Hochleistungs-Ein-/Ausgabe Michael Kuhn 2019-04-09 Wissenschaliches Rechnen Fachbereich Informatik Universität Hamburg Dateisysteme Orientierung Grundlagen Struktur Beispiel: ext4 Object Stores Datenstrukturen Leistungsbewertung Ausblick und Zusammenfassung Michael Kuhn Dateisysteme 1/49 E/A-Schichten Orientierung Anwendung Bibliotheken Paralleles verteiltes Dateisystem Dateisystem Optimierungen Datenreduktion Leistungsanalyse Speichergerät/-verbund Abbildung 1: E/A-Schichten und orthogonale Themen Michael Kuhn Dateisysteme 2/49 Aufgabe Grundlagen 1. Strukturierung • Üblicherweise hierarchische Organisation • Auf Basis von Dateien und Verzeichnissen • Andere Ansätze: Tagging, Queries etc. 2. Verwaltung von Daten und Metadaten • Blockallokation und -verwaltung • Zugri über Datei- und Verzeichnisnamen • Zugrisrechte, Zeitstempel etc. • Dateisysteme nutzen ein darunter liegendes Speichergerät • Kann auch durch Speicherverbund bereitgestellt werden • Logical Volume Manager (LVM) und/oder mdadm unter Linux Michael Kuhn Dateisysteme 3/49 Beispiele Grundlagen • Linux: tmpfs, ext4, XFS, btrfs, ZFS • Windows: FAT, exFAT, NTFS • OS X: HFS+, APFS • Universal: ISO9660, UDF • Pseudo: sysfs, proc Michael Kuhn Dateisysteme 4/49 Beispiele... Grundlagen • Netzwerk: NFS, AFS, Samba • Kryptographisch: EncFS, eCryptfs • Parallel verteilt: Spectrum Scale, Lustre, OrangeFS, CephFS, GlusterFS • Setzen häufig auf darunterliegenden Dateisystemen auf Michael Kuhn Dateisysteme 5/49 E/A-Schnittstellen Grundlagen • Anfragen werden über E/A-Schnittstellen realisiert • Schnittstellen für unterschiedliche Abstraktionsebenen • Weiterleitung an das eigentliche Dateisystem • Low-Level-Funktionalität • POSIX, MPI-IO • High-Level-Funktionalität • HDF, NetCDF, ADIOS Michael Kuhn Dateisysteme 6/49 E/A-Operationen Grundlagen 1 fd = open("/path/to/file", O_RDWR | O_CREAT | O_TRUNC, 2 S_IRUSR | S_IWUSR); 3 rv = close(fd); 4 rv = unlink("/path/to/file"); Listing 1: E/A über POSIX-Funktionen • Mit open können auch Dateien erstellt werden • Viele mögliche Flags und Modi • Initialer Zugri über Pfad • Danach über File Descriptor (bis auf einige Ausnahmen) • Alle Funktionen liefern einen Rückgabewert • Bei Fehlern sollte errno überprü werden Michael Kuhn Dateisysteme 7/49 E/A-Operationen... Grundlagen 1 nb = write(fd, data, sizeof(data)); Listing 2: E/A über POSIX-Funktionen • write liefert die Anzahl der geschriebenen Bytes zurück • Muss nicht notwendigerweise der übergebenen Größe entsprechen (Fehlerbehandlung!) • write verändert intern den Dateizeiger (alternativ: pwrite) • Funktionen befinden sich in der libc • Diese führt System Calls durch Michael Kuhn Dateisysteme 8/49 VFS Grundlagen • Virtual File System (Switch) • Zentrale Dateisystemkomponente im Kernel • Standardisiertes Interface für alle Dateisysteme (POSIX) • Gibt Dateisystemstruktur und -schnittstelle größtenteils vor • Leitet Anfragen der Anwendungen an das entsprechende Dateisystem weiter • Basierend auf dem Mountpoint • Ermöglicht die Unterstützung unterschiedlichster Dateisysteme • Anwendungen bleiben durch POSIX trotzdem portabel Michael Kuhn Dateisysteme 9/49 VFS... [3] Grundlagen mmap (anonymous pages) Applications (processes) malloc ... stat(2) open(2) read(2) write(2) chmod(2) VFS Block-based FS Network FS Pseudo FS Special ext2 ext3 ext4 xfs NFS coda purpose FS Direct I/O proc sysfs Page btrfs ifs iso9660 smbfs ... (O_DIRECT) pipefs futexfs tmpfs ramfs cache gfs ocfs ... ceph usbfs ... devtmpfs Stackable FS ecryptfs overlayfs unionfs FUSE userspace (e.g. sshfs) network stackable (optional) struct bio - sector on disk BIOs (block I/Os) Devices on top of “normal” BIOs (block I/Os) - sector cnt block devices - bio_vec cnt drbd LVM - bio_vec index device mapper mdraid - bio_vec list dm-crypt dm-mirror ... dm-cache dm-thin bcache dm-raid dm-delay The Linux Storage Stack Diagram http://www.thomas-krenn.com/en/wiki/Linux_Storage_Stack_Diagram Created by Werner Fischer and Georg Schönberger License: CC-BY-SA 3.0, see http://creativecommons.org/licenses/by-sa/3.0/ Michael Kuhn Dateisysteme 10 / 49 VFS... [3] Grundlagen The Linux Storage Stack Diagram version 4.10, 2017-03-10 outlines the Linux storage stack as of Kernel version 4.10 Virtual Host Virtual FireWire Fibre Channel Fibre over Ethernet ISCSI Channel Fibre USB mmap (anonymous pages) Applications (processes) LIO malloc vfs_writev, vfs_readv, ... ... stat(2) open(2) read(2) write(2) chmod(2) VFS tcm_fc sbp_target tcm_usb_gadget tcm_vhost tcm_qla2xxx iscsi_target_mod Block-based FS Network FS Pseudo FS Special ext2 ext3 ext4 xfs NFS coda proc purpose FS target_core_mod Direct I/O sysfs Page btrfs ifs iso9660 smbfs ... (O_DIRECT) pipefs futexfs tmpfs ramfs cache target_core_file gfs ocfs ... ceph usbfs ... devtmpfs Stackable FS target_core_iblock ecryptfs overlayfs unionfs FUSE userspace (e.g. sshfs) target_core_pscsi target_core_user network stackable (optional) struct bio - sector on disk BIOs (block I/Os) Devices on top of “normal” BIOs (block I/Os) - sector cnt block devices - bio_vec cnt drbd LVM - bio_vec index device mapper mdraid - bio_vec list dm-crypt dm-mirror ... dm-cache dm-thin bcache dm-raid dm-delay userspace BIOs BIOs Block Layer BIOs I/O scheduler blkmq Maps BIOs to requests multi queue hooked in device drivers noop Software (they hook in like stacked cfq ... queues devices do) deadline Hardware Hardware dispatch ... dispatch queue queues Request Request BIO based drivers based drivers based drivers Request-based device mapper targets dm-multipath SCSI mid layer sysfs scsi-mq /dev/zram* /dev/rbd* /dev/mmcblk*p* /dev/nullb* /dev/vd* /dev/rssd* /dev/skd* (transport attributes) SCSI upper level drivers /dev/sda /dev/sd* ... /dev/ubiblock* /dev/nbd* /dev/loop* /dev/nvme*n* /dev/rsxx* Transport classes scsi_transport_fc /dev/sr* /dev/st* zram ubi rbd nbd mmc loop null_blk virtio_blk mtip32xx nvme skd rsxx scsi_transport_sas scsi_transport_... network memory SCSI low level drivers libata megaraid_sas qla2xxx pm8001 iscsi_tcp virtio_scsi ufs ... ahci ata_piix ... aacraid lpfc mpt3sas vmw_pvscsi network HDD SSD DVD LSI Qlogic PMC-Sierra Micron nvme stec para-virtualized virtio_pci mobile device drive RAID HBA HBA SCSI flash memory PCIe card device device Adaptec Emulex LSI 12Gbs VMware's SD-/MMC-Card RAID HBA SAS HBA para-virtualized IBM flash Physical devices SCSI adapter The Linux Storage Stack Diagram http://www.thomas-krenn.com/en/wiki/Linux_Storage_Stack_Diagram Created by Werner Fischer and Georg Schönberger License: CC-BY-SA 3.0, see http://creativecommons.org/licenses/by-sa/3.0/ Michael Kuhn Dateisysteme 11 / 49 Dateisystemobjekte Struktur • Unterscheidung in Benutzer- und Systemsicht • Benutzer sehen Dateien und Verzeichnisse mit Metadaten • Dateien bestehen aus Bytes, Verzeichnisse enthalten Dateien und Verzeichnisse • Das System verwaltet Interna • Fasst mehrere Blöcke zu Dateien zusammen etc. • Inodes • Eigentliche Basisobjekte des Dateisystems • Jeder Datei und jedem Verzeichnis ist ein Inode zugeordnet (siehe stat) • Enthalten Metadaten • Sowohl für Benutzer sichtbare als auch interne • Üblicherweise eindeutige IDs und fixe Größe Michael Kuhn Dateisysteme 12 / 49 Dateisystemobjekte... Struktur • Dateien • Enthalten Daten in Form eines Byte-Arrays • POSIX macht keine weitergehenden Vorgaben • Können gelesen/geschrieben werden (explizit) • Können in den Speicher gemappt werden (implizit) • Verzeichnisse • Zur Organisation des Namensraumes • Können Dateien und weitere Verzeichnisse enthalten • Aus Benutzersicht eine Liste • Intern häufig Baumstrukturen Michael Kuhn Dateisysteme 13 / 49 Dateien Struktur 1 nb = pwrite(fd, data, sizeof(data), 42); 2 nb = pread(fd, data, sizeof(data), 42); Listing 3: Expliziter Zugri • pwrite und pread verhalten sich wie write bzw. read • Explizite Angabe des Osets und damit threadsicher • Zugri über File Descriptor • Kann von mehreren Threads parallel genutzt werden Michael Kuhn Dateisysteme 14 / 49 Dateien... Struktur 1 char* pt = mmap(NULL, file_size, PROT_READ | PROT_WRITE, 2 MAP_SHARED, fd, offset); 3 memcpy(pt + 42, data, sizeof(data)); 4 memcpy(data, pt + 42, sizeof(data)); 5 munmap(pt, FILE_SIZE); Listing 4: Impliziter Zugri • mmap erlaubt es eine Datei in den Speicher einzublenden • Datei wird an Adresse pt eingeblendet • Verschiedene Sichtbarkeitseinstellungen (shared vs. private) • Zugri wie auf andere Speicherobjekte • Z. B. via memcpy oder direkte Zuweisung Michael Kuhn Dateisysteme 15 / 49 Dateien... Struktur • Beide Zugrisarten haben jeweils Vor- und Nachteile • Beide Modi profitieren vom Caching durch das Betriebssystem • Expliziter Zugri • Vorteile: genaue Kontrolle über E/A, Möglichkeit direkter E/A • Nachteile: separate Puer notwendig, Kopiervorgänge zwischen Kernel- und Userspace • Impliziter Zugri • Vorteile: keine separaten Puer notwendig, eiziente E/A durch das Betriebssystem, keine unnötigen Kopiervorgänge, große Dateien können komplett gemappt werden • Nachteile: keine genaue Kontrolle, kompliziertere Fehlerbehandlung (Signale) Michael Kuhn Dateisysteme 16 / 49 Verzeichnisse Struktur Inode Größe Namenslänge Dateityp Name 23 10 2 2 . 24 11 3 2 .. 42 14 6 1 hello 42 14 6 1 world Abbildung 2: ext4-Verzeichniseintrag [1] • Traditionell lineares Array • Langsam, da über das komplette Array iteriert werden muss • Heutzutage eher Baumstrukturen • Deutlich komplexer, dafür geringere Zugriszeiten • Name wird nicht im Inode gespeichert • Mehrere Namen können auf denselben Inode zeigen Michael Kuhn Dateisysteme 17 / 49 Inodes Struktur Feldgröße Inhalt 2 Bytes Berechtigungen
Load more

Recommended publications
  • Efficient Implementation of Data Objects in the OSD+-Based Fusion
    Efficient Implementation of Data Objects in the OSD+-Based Fusion Parallel File System Juan Piernas(B) and Pilar Gonz´alez-F´erez Departamento de Ingenier´ıa y Tecnolog´ıa de Computadores, Universidad de Murcia, Murcia, Spain piernas,pilar @ditec.um.es { } Abstract. OSD+s are enhanced object-based storage devices (OSDs) able to deal with both data and metadata operations via data and direc- tory objects, respectively. So far, we have focused on designing and implementing efficient directory objects in OSD+s. This paper, however, presents our work on also supporting data objects, and describes how the coexistence of both kinds of objects in OSD+s is profited to efficiently implement data objects and to speed up some commonfile operations. We compare our OSD+-based Fusion Parallel File System (FPFS) with Lustre and OrangeFS. Results show that FPFS provides a performance up to 37 better than Lustre, and up to 95 better than OrangeFS, × × for metadata workloads. FPFS also provides 34% more bandwidth than OrangeFS for data workloads, and competes with Lustre for data writes. Results also show serious scalability problems in Lustre and OrangeFS. Keywords: FPFS OSD+ Data objects Lustre OrangeFS · · · · 1 Introduction File systems for HPC environment have traditionally used a cluster of data servers for achieving high rates in read and write operations, for providing fault tolerance and scalability, etc. However, due to a growing number offiles, and an increasing use of huge directories with millions or billions of entries accessed by thousands of processes at the same time [3,8,12], some of thesefile systems also utilize a cluster of specialized metadata servers [6,10,11] and have recently added support for distributed directories [7,10].
    [Show full text]
  • Globalfs: a Strongly Consistent Multi-Site File System
    GlobalFS: A Strongly Consistent Multi-Site File System Leandro Pacheco Raluca Halalai Valerio Schiavoni University of Lugano University of Neuchatelˆ University of Neuchatelˆ Fernando Pedone Etienne Riviere` Pascal Felber University of Lugano University of Neuchatelˆ University of Neuchatelˆ Abstract consistency, availability, and tolerance to partitions. Our goal is to ensure strongly consistent file system operations This paper introduces GlobalFS, a POSIX-compliant despite node failures, at the price of possibly reduced geographically distributed file system. GlobalFS builds availability in the event of a network partition. Weak on two fundamental building blocks, an atomic multicast consistency is suitable for domain-specific applications group communication abstraction and multiple instances of where programmers can anticipate and provide resolution a single-site data store. We define four execution modes and methods for conflicts, or work with last-writer-wins show how all file system operations can be implemented resolution methods. Our rationale is that for general-purpose with these modes while ensuring strong consistency and services such as a file system, strong consistency is more tolerating failures. We describe the GlobalFS prototype in appropriate as it is both more intuitive for the users and detail and report on an extensive performance assessment. does not require human intervention in case of conflicts. We have deployed GlobalFS across all EC2 regions and Strong consistency requires ordering commands across show that the system scales geographically, providing replicas, which needs coordination among nodes at performance comparable to other state-of-the-art distributed geographically distributed sites (i.e., regions). Designing file systems for local commands and allowing for strongly strongly consistent distributed systems that provide good consistent operations over the whole system.
    [Show full text]
  • Storage Systems and Input/Output 2018 Pre-Workshop Document
    Storage Systems and Input/Output 2018 Pre-Workshop Document Gaithersburg, Maryland September 19-20, 2018 Meeting Organizers Robert Ross (ANL) (lead organizer) Glenn Lockwood (LBL) Lee Ward (SNL) (co-lead) Kathryn Mohror (LLNL) Gary Grider (LANL) Bradley Settlemyer (LANL) Scott Klasky (ORNL) Pre-Workshop Document Contributors Philip Carns (ANL) Quincey Koziol (LBL) Matthew Wolf (ORNL) 1 Table of Contents 1 Table of Contents 2 2 Executive Summary 4 3 Introduction 5 4 Mission Drivers 7 4.1 Overview 7 4.2 Workload Characteristics 9 4.2.1 Common observations 9 4.2.2 An example: Adjoint-based sensitivity analysis 10 4.3 Input/Output Characteristics 11 4.4 Implications of In Situ Analysis on the SSIO Community 14 4.5 Data Organization and Archiving 15 4.6 Metadata and Provenance 18 4.7 Summary 20 5 Computer Science Challenges 21 5.1 Hardware/Software Architectures 21 5.1.1 Storage Media and Interfaces 21 5.1.2 Networks 21 5.1.3 Active Storage 22 5.1.4 Resilience 23 5.1.5 Understandability 24 5.1.6 Autonomics 25 5.1.7 Security 26 5.1.8 New Paradigms 27 5.2 Metadata, Name Spaces, and Provenance 28 5.2.1 Metadata 28 5.2.2 Namespaces 30 5.2.3 Provenance 30 5.3 Supporting Science Workflows - SAK 32 5.3.1 DOE Extreme Scale Use cases - SAK 33 5.3.2 Programming Model Integration - (Workflow Composition for on line workflows, and for offline workflows ) - MW 33 5.3.3 Workflows (Engine) - Provision and Placement MW 34 5.3.4 I/O Middleware and Libraries (Connectivity) - both on-and offline, (not or) 35 2 Storage Systems and Input/Output 2018 Pre-Workshop
    [Show full text]
  • HFAA: a Generic Socket API for Hadoop File Systems
    HFAA: A Generic Socket API for Hadoop File Systems Adam Yee Jeffrey Shafer University of the Pacific University of the Pacific Stockton, CA Stockton, CA [email protected] jshafer@pacific.edu ABSTRACT vices: one central NameNode and many DataNodes. The Hadoop is an open-source implementation of the MapReduce NameNode is responsible for maintaining the HDFS direc- programming model for distributed computing. Hadoop na- tory tree. Clients contact the NameNode in order to perform tively integrates with the Hadoop Distributed File System common file system operations, such as open, close, rename, (HDFS), a user-level file system. In this paper, we intro- and delete. The NameNode does not store HDFS data itself, duce the Hadoop Filesystem Agnostic API (HFAA) to allow but rather maintains a mapping between HDFS file name, Hadoop to integrate with any distributed file system over a list of blocks in the file, and the DataNode(s) on which TCP sockets. With this API, HDFS can be replaced by dis- those blocks are stored. tributed file systems such as PVFS, Ceph, Lustre, or others, thereby allowing direct comparisons in terms of performance Although HDFS stores file data in a distributed fashion, and scalability. Unlike previous attempts at augmenting file metadata is stored in the centralized NameNode service. Hadoop with new file systems, the socket API presented here While sufficient for small-scale clusters, this design prevents eliminates the need to customize Hadoop’s Java implementa- Hadoop from scaling beyond the resources of a single Name- tion, and instead moves the implementation responsibilities Node. Prior analysis of CPU and memory requirements for to the file system itself.
    [Show full text]
  • Maximizing the Performance of Scientific Data Transfer By
    Maximizing the Performance of Scientific Data Transfer by Optimizing the Interface Between Parallel File Systems and Advanced Research Networks Nicholas Millsa,∗, F. Alex Feltusb, Walter B. Ligon IIIa aHolcombe Department of Electrical and Computer Engineering, Clemson University, Clemson, SC bDepartment of Genetics and Biochemistry, Clemson University, Clemson, SC Abstract The large amount of time spent transferring experimental data in fields such as genomics is hampering the ability of scientists to generate new knowledge. Often, computer hardware is capable of faster transfers but sub-optimal transfer software and configurations are limiting performance. This work seeks to serve as a guide to identifying the optimal configuration for performing genomics data transfers. A wide variety of tests narrow in on the optimal data transfer parameters for parallel data streaming across Internet2 and between two CloudLab clusters loading real genomics data onto a parallel file system. The best throughput was found to occur with a configuration using GridFTP with at least 5 parallel TCP streams with a 16 MiB TCP socket buffer size to transfer to/from 4{8 BeeGFS parallel file system nodes connected by InfiniBand. Keywords: Software Defined Networking, High Throughput Computing, DNA sequence, Parallel Data Transfer, Parallel File System, Data Intensive Science 1. Introduction of over 2,500 human genomes from 26 populations have been determined (The 1000 Genomes Project [3]); genome Solving scientific problems on a high-performance com- sequences have been produced for 3,000 rice varieties from puting (HPC) cluster will happen faster by taking full ad- 89 countries (The 3000 Rice Genomes Project [4]). These vantage of specialized infrastructure such as parallel file raw datasets are but a few examples that aggregate into systems and advanced software-defined networks.
    [Show full text]
  • Locofs: a Loosely-Coupled Metadata Service for Distributed File Systems
    LocoFS: A Loosely-Coupled Metadata Service for Distributed File Systems Siyang Li∗ Youyou Lu Jiwu Shu† Tsinghua University Tsinghua University Tsinghua University [email protected] [email protected] [email protected] Tao Li Yang Hu University of Florida University of Texas, Dallas [email protected] [email protected] ABSTRACT 1 INTRODUCTION Key-Value stores provide scalable metadata service for distributed As clusters or data centers are moving from Petabyte level to Ex- file systems. However, the metadata’s organization itself, which is abyte level, distributed file systems are facing challenges in meta- organized using a directory tree structure, does not fit the key-value data scalability. The recent work IndexFS [38] uses hundreds of access pattern, thereby limiting the performance. To address this metadata servers to achieve high-performance metadata operations. issue, we propose a distributed file system with a loosely-coupled However, most of the recent active super computers only deploy metadata service, LocoFS, to bridge the performance gap between 1 to 4 metadata servers to reduce the complexity of management file system metadata and key-value stores. LocoFS is designed to and guarantee reliability. Besides, previous work [24, 39] has also decouple the dependencies between different kinds of metadata revealed that metadata operations consume more than half of all with two techniques. First, LocoFS decouples the directory content operations in file systems. The metadata service plays a major role and structure, which organizes file and directory index nodes ina in distributed file systems. It is important to support parallel pro- flat space while reversely indexing the directory entries.
    [Show full text]
  • A Persistent Storage Model for Extreme Computing Shuangyang Yang Louisiana State University and Agricultural and Mechanical College
    Louisiana State University LSU Digital Commons LSU Doctoral Dissertations Graduate School 2014 A Persistent Storage Model for Extreme Computing Shuangyang Yang Louisiana State University and Agricultural and Mechanical College Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_dissertations Part of the Computer Sciences Commons Recommended Citation Yang, Shuangyang, "A Persistent Storage Model for Extreme Computing" (2014). LSU Doctoral Dissertations. 2910. https://digitalcommons.lsu.edu/gradschool_dissertations/2910 This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Doctoral Dissertations by an authorized graduate school editor of LSU Digital Commons. For more information, please [email protected]. A PERSISTENT STORAGE MODEL FOR EXTREME COMPUTING A Dissertation Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy in The Department of Computer Science by Shuangyang Yang B.S., Zhejiang University, 2006 M.S., University of Dayton, 2008 December 2014 Copyright © 2014 Shuangyang Yang All rights reserved ii Dedicated to my wife Miao Yu and our daughter Emily. iii Acknowledgments This dissertation would not be possible without several contributions. It is a great pleasure to thank Dr. Hartmut Kaiser @ Louisiana State University, Dr. Walter B. Ligon III @ Clemson University and Dr. Maciej Brodowicz @ Indiana University for their ongoing help and support. It is a pleasure also to thank Dr. Bijaya B. Karki, Dr. Konstantin Busch, Dr. Supratik Mukhopadhyay at Louisiana State University for being my committee members and Dr.
    [Show full text]
  • The Ceph Distributed Storage System
    the ceph distributed storage system sage weil msst – april 17, 2012 outline ● why you should care ● what is it, what it does ● how it works, how you can use it ● architecture ● objects and data placement ● file system ● big data, cloud ● current status, roadmap ● who we are, why we do this why should you care about another storage system? requirements, time, money storage requirements ● scale ● terabytes, petabytes, exabytes ● heterogeneous hardware ● reliability and fault tolerance ● diverse storage needs ● object storage ● block devices ● shared file system (POSIX, coherent caches) ● structured data time ● ease of administration ● no manual data migration, load balancing ● painless scaling ● expansion and contraction ● seamless migration money ● low cost per gigabyte ● no vendor lock-in ● software solution ● commodity hardware ● open source what is ceph? unified storage system ● objects ● small or large ● multi-protocol Netflix VM Hadoop ● block devices radosgw RBD Ceph DFS ● snapshots, cloning RADOS ● files ● cache coherent ● snapshots ● usage accounting open source ● LGPLv2 ● copyleft ● free to link to proprietary code ● no copyright assignment ● no dual licensing ● no “enterprise-only” feature set ● active community ● commercial support distributed storage system ● data center (not geo) scale ● 10s to 10,000s of machines ● terabytes to exabytes ● fault tolerant ● no SPoF ● commodity hardware – ethernet, SATA/SAS, HDD/SSD – RAID, SAN probably a waste of time, power, and money architecture ● monitors (ceph-mon) ● 1s-10s, paxos ● lightweight
    [Show full text]
  • Introduction to Distributed File Systems. Orangefs Experience 0.05 [Width=0.4]Lvee-Logo-Winter
    Introduction to distributed file systems. OrangeFS experience Andrew Savchenko NRNU MEPhI, Moscow, Russia 16 February 2013 . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Outline 1. Introduction 2. Species of distributed file systems 3. Behind the curtain 4. OrangeFS 5. Summary . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Introduction Why one needs a non-local file system? • a large data storage • a high performance data storage • redundant and highly available solutions There are dozens of them: 72 only on wiki[1], more IRL. Focus on free software solutions. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Introduction Why one needs a non-local file system? • a large data storage • a high performance data storage • redundant and highly available solutions There are dozens of them: 72 only on wiki[1], more IRL. Focus on free software solutions. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Species of distributed file systems Network Clustered Distributed Terminology is ambiguous!. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Network file systems client1 client2 clientN server storage A single server (or at least an appearance) and multiple network clients. Examples: NFS, CIFS. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Clustered file systems server1 server2 serverN storage Servers sharing the same local storage (usually SAN[2] at block level). shared storage architecture. Examples: GFS2[3], OCFS2[4]. .. .. .. .. .
    [Show full text]
  • Týrfs: Increasing Small Files Access Performance with Dynamic Metadata Replication Pierre Matri, María Pérez, Alexandru Costan, Gabriel Antoniu
    TýrFS: Increasing Small Files Access Performance with Dynamic Metadata Replication Pierre Matri, María Pérez, Alexandru Costan, Gabriel Antoniu To cite this version: Pierre Matri, María Pérez, Alexandru Costan, Gabriel Antoniu. TýrFS: Increasing Small Files Access Performance with Dynamic Metadata Replication. CCGRID 2018 - 18th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, May 2018, Washington, United States. pp.452- 461, 10.1109/CCGRID.2018.00072. hal-01892691 HAL Id: hal-01892691 https://hal.archives-ouvertes.fr/hal-01892691 Submitted on 10 Oct 2018 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. TyrFS:´ Increasing Small Files Access Performance with Dynamic Metadata Replication Pierre Matri∗, Mar´ıa S. Perez´ ∗, Alexandru Costanyz, Gabriel Antoniuz ∗Universidad Politecnica´ de Madrid, Madrid, Spain, fpmatri, mperezg@fi.upm.es yIRISA / INSA Rennes, Rennes, France, [email protected] zInria, Rennes, France, falexandru.costan, [email protected] Abstract—Small files are known to pose major performance We advocate that a different file system architecture is challenges for file systems. Yet, such workloads are increasingly necessary to reduce the cost of metadata management for common in a number of Big Data Analytics workflows or large- such workloads involving many small files.
    [Show full text]
  • Filesystems in IO500 3% Gridscaler Datawarp 2% Wekaio Matrix 2% Total Submissions Across All Lists: 147 5% SUSE Enterprise Storage 1% Lustre 29%
    Lustre and IO-500 The Lustre® parallel file system has a long tradition in appearing as the file system of choice for many of the Top500 [1] supercomputer systems. A new IO-centric benchmarking project known as the IO-500 [2] has been created. The IO-500 was developed as a community led effort to highlight the importance of I/O in HPC systems as well as drive storage system vendors to look at I/O patterns other than block-aligned large I/O requests. Lustre features prominently in the IO-500 with 29% of the overall submission listing Lustre and six of the top ten ranked systems are using Lustre from the September 2019 Supercomputing edition of the list. Submitting to the IO-500 challenge helps to bring attention to Lustre platform and promote the use of open software. Andreas Dilger provided a guide to understanding the IO-500 benchmark and tuning changes as part of the Lustre Users Group 2019 conference. [3] Please take a look at the presentation for ideas on how to improve your Lustre file system performance. The IO-500 also offers a simplified submission known as the “10-node Challenge”, requiring only 10 client nodes to perform the test. DAOS GPFS 1% Filesystems in IO500 3% GRIDScaler DataWarp 2% WekaIO Matrix 2% Total submissions across all lists: 147 5% SUSE Enterprise Storage 1% Lustre 29% Spectrum Scale 12% Scale-out NAS 7% BeeGFS Scale out NFS IME 20% 4% OrangeFS 7% 1% NFS 1% HPE XFS CephFS 2% 3% Courtesy of Oak Ridge National Laboratory About OpenSFS Open Scalable File Systems, Inc.
    [Show full text]
  • Metadata Management in Distributed File Systems
    Metadata Management in Distributed File Systems University of Oulu Faculty of Information Technology and Electrical Engineering Bachelor’s Thesis Joni Laitala 6.9.2017 2 Abstract The purpose of this research has been to study the architectures of popular distributed file systems used in cloud computing, with a focus on their metadata management, in order to identify differences between and issues within varying designs from the metadata perspective. File system and metadata concepts are briefly introduced before the comparisons are made. Keywords big data, distributed file system, distributed storage system, metadata management Supervisor Dr., University Lecturer Ari Vesanen 3 Contents Abstract ............................................................................................................................. 2 Contents ............................................................................................................................ 3 1. Introduction .................................................................................................................. 4 2. File Systems ................................................................................................................. 5 2.1 Disk File Systems ........................................................................................ 5 2.2 Cloud Distributed File Systems ................................................................... 5 2.3 Metadata .....................................................................................................
    [Show full text]