Content deleted Content added
m Reverted edit by 154.133.103.135 (talk) to last version by JCW-CleanerBot |
|||
| (30 intermediate revisions by 15 users not shown) | |||
Line 1:
{{Short description|File system that allows many clients to have access}}
A '''distributed file system for cloud''' is a [[w:file system|file system]] that allows many clients to have access to data and supports operations (create, delete, modify, read, write) on that data. Each data file may be partitioned into several parts called [[Chunk (information)|chunks]]. Each chunk may be stored on different remote machines, facilitating the parallel execution of applications. Typically, data is stored in files in a [[Hierarchical tree structure|hierarchical tree]], where the nodes represent directories. There are several ways to share files in a distributed architecture: each solution must be suitable for a certain type of application, depending on how complex the application is. Meanwhile, the security of the system must be ensured. [[w:Confidentiality|Confidentiality]], [[w:Availability|availability]] and [[w:Integrity|integrity]] are the main keys for a secure system.
Line 6 ⟶ 7:
== Overview ==
=== History ===
Today, there are many implementations of distributed file systems. The first file servers were developed by researchers in the 1970s. Sun Microsystem's [[Network File System]] became available in the 1980s. Before that, people who wanted to share files used the [[sneakernet]] method, physically transporting files on storage media from place to place. Once computer networks started to proliferate, it became obvious that the existing file systems had many limitations and were unsuitable for multi-user environments. Users initially used [[FTP]] to share files.<ref>{{harvnb|Sun microsystem|p=1}}</ref> FTP first ran on the [[PDP-10]] at the end of 1973. Even with FTP, files needed to be copied from the source computer onto a server and then from the server onto the destination computer. Users were required to know the physical addresses of all computers involved with the file sharing.<ref>{{harvnb|
=== Supporting techniques ===
Line 22:
=== Client-server architecture ===
[[Network File System]] (NFS) uses a [[client-server architecture]], which allows sharing of files between a number of machines on a network as if they were located locally, providing a standardized view. The NFS protocol allows heterogeneous clients' processes, probably running on different machines and under different operating systems, to access files on a distant server, ignoring the actual ___location of files. Relying on a single server results in the NFS protocol suffering from potentially low availability and poor scalability. Using multiple servers does not solve the availability problem since each server is working independently.<ref>{{harvnb|Di Sano| Di Stefano|Morana|Zito|2012|p=2}}</ref> The model of NFS is a remote file service. This model is also called the remote access model, which is in contrast with the upload/download model:
* Remote access model: Provides transparency, the client has access to a file. He
* Upload/download model: The client can access the file only locally. It means that the client has to download the file, make modifications, and upload it again, to be used by others' clients.
Line 32:
==== Design principles ====
===== Goals =====
[[Google File System]] (GFS) and [[Hadoop Distributed File System]] (HDFS) are specifically built for handling [[batch processing]] on very large data sets.
Line 43 ⟶ 42:
===== Load balancing =====
[[Load balancing (computing)|Load balancing]] is essential for efficient operation in distributed environments. It means distributing work among different servers,<ref>{{harvnb|Kai|Dayang|Hui|Yintang|2013|p=23}}</ref> fairly, in order to get more work done in the same amount of time and to serve clients faster. In a system containing N chunkservers in a cloud (N being 1000, 10000, or more), where a certain number of files are stored, each file is split into several parts or chunks of fixed size (for example, 64 megabytes), the load of each chunkserver being proportional to the number of chunks hosted by the server.<ref name="ReferenceA">{{harvnb|Hsiao|Chung|Shen|Chao|2013|p=2}}</ref> In a load-balanced cloud, resources can be efficiently used while maximizing the performance of MapReduce-based applications.
===== Load rebalancing =====
In a cloud computing environment, failure is the norm,<ref>{{harvnb|Hsiao|Chung|Shen|Chao|2013|p=952}}</ref><ref>{{harvnb|Ghemawat|Gobioff|Leung|2003|p=1}}</ref> and chunkservers may be upgraded, replaced, and added to the system. Files can also be dynamically created, deleted, and appended. That leads to load imbalance in a distributed file system, meaning that the file chunks are not distributed equitably between the servers.
Distributed file systems in clouds such as GFS and HDFS rely on central or master servers or nodes (Master for GFS and NameNode for HDFS) to manage the metadata and the load balancing. The master rebalances replicas periodically: data must be moved from one DataNode/chunkserver to another if free space on the first server falls below a certain threshold.<ref>{{harvnb|Ghemawat|Gobioff|Leung|2003|p=8}}</ref> However, this centralized approach can become a bottleneck for those master servers, if they become unable to manage a large number of file accesses, as it increases their already heavy loads. The load rebalance problem is [[w:NP-hard|NP-hard]].<ref>{{harvnb|Hsiao|Chung|Shen|Chao|2013|p=953}}</ref>
In order to get a large number of chunkservers to work in collaboration, and to solve the problem of load balancing in distributed file systems, several approaches have been proposed, such as reallocating file chunks so that the chunks can be distributed as uniformly as possible while reducing the movement cost as much as possible.<ref name="ReferenceA" />
==== Google file system ====
{{
===== Description =====
Line 63 ⟶ 60:
The master server running in dedicated node is responsible for coordinating storage resources and managing files's [[metadata]] (the equivalent of, for example, inodes in classical file systems).<ref name="Krzyzanowski_p2">{{harvnb|Krzyzanowski|2012|p=2}}</ref>
Each file is split
The master maintains all of the files's metadata, including file names, directories, and the mapping of files to the list of chunks that contain each file's data. The metadata is kept in the master server's main memory, along with the mapping of files to chunks. Updates to this data are logged to an operation log on disk. This operation log is replicated onto remote machines. When the log
===== Fault tolerance =====
Line 85 ⟶ 82:
==== Hadoop distributed file system ====
{{
{{abbr|HDFS |Hadoop Distributed File System}}, developed by the [[Apache Software Foundation]], is a distributed file system designed to hold very large amounts of data (terabytes or even petabytes). Its architecture is similar to GFS, i.e. a
The design concept of Hadoop is informed by Google's, with Google File System, Google MapReduce and [[Bigtable]], being implemented by Hadoop Distributed File System (HDFS), Hadoop MapReduce, and Hadoop Base (HBase) respectively.<ref>{{harvnb|Fan-Hsun|Chi-Yuan| Li-Der| Han-Chieh|2012|p=2}}</ref> Like GFS, HDFS is suited for scenarios with write-once-read-many file access, and supports file appends and truncates in lieu of random reads and writes to simplify data coherency issues.<ref>{{Cite web | url=http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html#Assumptions_and_Goals | title=Apache Hadoop 2.9.2 – HDFS Architecture}}</ref>
Line 103 ⟶ 100:
Distributed file systems can be optimized for different purposes. Some, such as those designed for internet services, including GFS, are optimized for scalability. Other designs for distributed file systems support performance-intensive applications usually executed in parallel.<ref>{{harvnb|Soares| Dantas†|de Macedo|Bauer|2013|p=158}}</ref> Some examples include: [[MapR FS|MapR File System]] (MapR-FS), [[Ceph (storage)|Ceph-FS]], [[BeeGFS|Fraunhofer File System (BeeGFS)]], [[Lustre (file system)|Lustre File System]], [[IBM General Parallel File System]] (GPFS), and [[Parallel Virtual File System]].
MapR-FS is a distributed file system that is the basis of the MapR Converged Platform, with capabilities for distributed file storage, a NoSQL database with multiple APIs, and an integrated message streaming system. MapR-FS is optimized for scalability, performance, reliability, and availability. Its file storage capability is compatible with the Apache Hadoop Distributed File System (HDFS) API but with several design characteristics that distinguish it from HDFS. Among the most notable differences are that MapR-FS is a fully read/write filesystem with metadata for files and directories distributed across the namespace, so there is no NameNode.<ref name="mapr-productivity">{{cite web|last1=Perez|first1=Nicolas|title=How MapR improves our productivity and simplifies our design|url=https://medium.com/@anicolaspp/how-mapr-improves-our-productivity-and-simplify-our-design-2d777ab53120#.mvr6mmydr|website
Ceph-FS is a distributed file system that provides excellent performance and reliability.<ref>{{harvnb|Weil|Brandt|Miller|Long|2006|p=307}}</ref> It answers the challenges of dealing with huge files and directories, coordinating the activity of thousands of disks, providing parallel access to metadata on a massive scale, manipulating both scientific and general-purpose workloads, authenticating and encrypting on a large scale, and increasing or decreasing dynamically due to frequent device decommissioning, device failures, and cluster expansions.<ref>{{harvnb|Maltzahn|Molina-Estolano|Khurana|Nelson|2010|p=39}}</ref>
Line 114 ⟶ 111:
High performance of distributed file systems requires efficient communication between computing nodes and fast access to the storage systems. Operations such as open, close, read, write, send, and receive need to be fast, to ensure that performance. For example, each read or write request accesses disk storage, which introduces seek, rotational, and network latencies.<ref>{{harvnb|Upadhyaya|Azimov|Doan|Choi|2008|p=400}}</ref>
The data communication (send/receive) operations transfer data from the application buffer to the machine kernel, [[Transmission Control Protocol|TCP]] controlling the process and being implemented in the kernel. However, in case of network congestion or errors, TCP may not send the data directly. While transferring data from a buffer in the [[
Choosing the buffer-size, for file reading and writing, or file sending and receiving, is done at the application level. The buffer is maintained using a [[Linked list|circular linked list]].<ref>{{harvnb|Upadhyaya|Azimov|Doan|Choi|2008|p=401}}</ref> It consists of a set of BufferNodes. Each BufferNode has a DataField. The DataField contains the data and a pointer called NextBufferNode that points to the next BufferNode. To find the current position, two [[Pointer (computer programming)|pointers]] are used: CurrentBufferNode and EndBufferNode, that represent the position in the BufferNode for the last write and read positions.
Line 188 ⟶ 185:
== Bibliography ==
* {{cite book
| last1 = Andrew
| first1 = S.Tanenbaum | last2 = Maarten
| first2 = Van Steen | year = 2006
| title = Distributed systems principles and paradigms
| url = http://net.pku.edu.cn/~course/cs501/2011/resource/2006-Book-distributed%20systems%20principles%20and%20paradigms%202nd%20edition.pdf
| access-date = 2014-01-10
| archive-date = 2013-08-20
| archive-url = https://web.archive.org/web/20130820190519/http://net.pku.edu.cn/~course/cs501/2011/resource/2006-Book-distributed%20systems%20principles%20and%20paradigms%202nd%20edition.pdf
| url-status = dead
}}
* {{cite web
| first = Fabio |last = Kon
| title = Distributed File Systems Past, Present and Future: A Distributed File System for 2006
| url = https://www.researchgate.net/publication/2439179
| year = 1996
| website = [[ResearchGate]]
}}
* {{cite web
Line 212 ⟶ 216:
* {{cite web
| last1 = Jacobi
| first1 = Tim-Daniel
| last2 = Lingemann
| first2 = Jan
| url = http://wr.informatik.uni-hamburg.de/_media/research/labs/2012/2012-10-tim-daniel_jacobi_jan_lingemann-evaluation_of_distributed_file_systems-report.pdf
| title = Evaluation of Distributed File Systems
| access-date = 2014-01-24
| archive-date = 2014-02-03
| archive-url = https://web.archive.org/web/20140203140412/http://wr.informatik.uni-hamburg.de/_media/research/labs/2012/2012-10-tim-daniel_jacobi_jan_lingemann-evaluation_of_distributed_file_systems-report.pdf
| url-status = dead
}}
# Architecture, structure, and design:
#* {{cite book
Line 232 ⟶ 240:
| year = 2012
| doi = 10.1109/ClusterW.2012.27
| s2cid = 12430485
| chapter = A Novel Scalable Architecture of Cloud Storage System for Small Files Based on P2P
Line 243 ⟶ 250:
| year = 2013
| doi = 10.1109/CTS.2013.6567222
| s2cid = 45293053
| pages = 155–161
Line 255 ⟶ 261:
| year = 2012
| url = http://www.cs.rutgers.edu/~pxk/417/notes/16-dfs.pdf
| access-date = 2013-12-27
| archive-date = 2013-12-27
| archive-url = https://web.archive.org/web/20131227152320/http://www.cs.rutgers.edu/~pxk/417/notes/16-dfs.pdf
| url-status = dead
}}
#* {{cite conference
| last1 = Kobayashi | first1 = K
Line 264 ⟶ 274:
| title = The Gfarm File System on Compute Clouds
| conference = Parallel and Distributed Processing Workshops and Phd Forum (IPDPSW), 2011 IEEE International Symposium on
| conference-url =
| doi = 10.1109/IPDPS.2011.255
}}
#* {{cite book
Line 274 ⟶ 283:
| year = 2012
| doi = 10.1109/ICAICT.2012.6398489
| s2cid = 6113112
| pages = 1–5
Line 290 ⟶ 298:
| first4 =Yu-Chang
| title = Load Rebalancing for Distributed File Systems in Clouds
|
| year = 2013
| doi = 10.1109/TPDS.2012.196
| s2cid = 11271386
| pages = 951–962
Line 311 ⟶ 318:
| year = 2013
| doi = 10.1109/INCoS.2013.14
| s2cid = 14821266
| pages = 23–29
Line 333 ⟶ 339:
| year = 2008
| doi = 10.1109/NCM.2008.164
| s2cid = 18933772
| pages = 400–405
Line 351 ⟶ 356:
| year = 2013
| doi = 10.1109/WETICE.2013.12
| s2cid = 6155753
| pages = 158–163
Line 363 ⟶ 367:
| year = 2012
| doi = 10.1109/ICAICT.2012.6398484
| s2cid = 16674289
| pages = 1–3
| chapter = Distributed file system as a basis of data-intensive computing
Line 375 ⟶ 378:
| year = 2003
| url = https://www.kernel.org/doc/ols/2003/ols2003-pages-380-386.pdf
| pages = 400–407
}}
#* {{cite journal
|
|first1 = Terry
|
|first2 = Alice
|last3 = Yates
|first3 = R. Kim
|
|
|
|year = 2000
|access-date = 2014-01-24
|archive-date = 2013-02-26
|archive-url = https://web.archive.org/web/20130226053255/https://computing.llnl.gov/code/sio/GPFS_performance.pdf
|url-status = dead
}}
#* {{cite
|
|
|
|
|
|
|
|
|
|
|
|conference = Proceedings of the 7th Conference on Operating Systems Design and Implementation (OSDI '06)
|access-date = 2014-01-24
|archive-date = 2012-03-09
|archive-url = https://web.archive.org/web/20120309021423/http://www.ssrc.ucsc.edu/Papers/weil-osdi06.pdf
|url-status = dead
}}
#* {{cite
| last1 = Maltzahn
| first1 = Carlos
Line 415 ⟶ 424:
| first4= Alex J.
| last5 = Brandt
| first5= Scott A.
| last6=Weil
| first6=Sage
| title =Ceph as a scalable alternative to the Hadoop Distributed FileSystem
| year = 2010
Line 434 ⟶ 443:
| year = 2003
| doi = 10.1109/MASS.2003.1194865
| pages = 290–298
| chapter = Efficient metadata management in large distributed storage systems
| isbn = 978-0-7695-1914-2
| citeseerx = 10.1.1.13.2537
| s2cid = 5548463
}}
#* {{cite journal
Line 453 ⟶ 462:
| url =https://www.cs.cmu.edu/~garth/CACM/CACM00-p37-gibson.pdf
| doi=10.1145/353360.353362
| s2cid = 207644891
}}
#* {{cite arXiv
| last1 = Yee
| first1 = Tin Tin
Line 489 ⟶ 499:
| year = 2011
| doi = 10.1109/SWS.2011.6101263
| s2cid = 14791637
| pages = 16–20
| chapter = A carrier-grade service-oriented file storage architecture for cloud computing
Line 508 ⟶ 517:
| chapter = The Google file system
| isbn = 978-1-58113-757-6
| s2cid =221261373
}}
# Security
Line 521 ⟶ 530:
| year = 2009
| doi = 10.1109/I-SPAN.2009.150
| pages = 4–16
| chapter = High-Performance Cloud Computing: A View of Scientific Applications
| isbn = 978-1-4244-5403-7
| arxiv = 0910.1979
| s2cid = 1810240
}}
#* {{cite book
Line 552 ⟶ 561:
| isbn = 978-1-4503-1004-8
| citeseerx = 10.1.1.225.8007
| s2cid = 12274859
}}
#* {{cite book
Line 562 ⟶ 572:
| year = 2012
| doi = 10.1109/MIC.2012.6273264
| s2cid = 40685246
| pages = 327–331
| chapter = PsFS: A high-throughput parallel file system for secure Cloud Storage system
Line 577 ⟶ 586:
| last4 = Xue
| first4 = Lan
| title = Efficient Metadata Management in Large Distributed Storage Systems
| periodical = 11th NASA Goddard Conference on Mass Storage Systems and Technologies, San Diego, CA
| year = 2003
| url = http://www.ssrc.ucsc.edu/Papers/brandt-mss03.pdf
| access-date = 2013-12-27
| archive-date = 2013-08-22
| archive-url = https://web.archive.org/web/20130822213717/http://www.ssrc.ucsc.edu/Papers/brandt-mss03.pdf
| url-status = dead
}}
#* {{cite journal
| author = Lori M. Kaufman
| s2cid = 16233643
| title =Data Security in the World of Cloud Computing
|
| year = 2009
| doi = 10.1109/MSP.2009.87
Line 601 ⟶ 613:
| last3 = Oprea
| first3 =Alina
| title = Proceedings of the 16th ACM conference on Computer and communications security
| chapter = HAIL: A high-availability and integrity layer for cloud storage
| s2cid = 207176701
| year = 2009
| doi = 10.1145/1653662.1653686
Line 632 ⟶ 644:
| year = 2012
| doi = 10.1109/Grid.2012.17
| s2cid = 10778240
| pages = 12–21
| chapter = A Distributed Cache for Hadoop Distributed File System in Real-Time Cloud Services
Line 652 ⟶ 663:
| year = 2012
| doi = 10.1109/SC.Companion.2012.103
| s2cid = 5554936
| pages = 753–759
| chapter = Integrating High Performance File Systems in a Cloud Computing Environment
Line 670 ⟶ 680:
| year = 2012
| doi = 10.1109/ISPACS.2012.6473485
| s2cid = 18260943
| pages = 227–232
| chapter = Implement a reliable and secure cloud distributed file system
Line 688 ⟶ 697:
| year = 2012
| doi = 10.1109/WETICE.2012.104
| s2cid = 19798809
| pages = 173–178
| chapter = File System As-a-Service: Providing Transient and Consistent Views of Files to Cooperating Applications in Clouds
Line 715 ⟶ 723:
| year = 2008
| url = http://www.pewinternet.org/~/media//Files/Reports/2008/PIP_Cloud.Memo.pdf.pdf
| access-date = 2013-12-27
| archive-date = 2013-07-12
| archive-url = https://web.archive.org/web/20130712182757/http://www.pewinternet.org/~/media//Files/Reports/2008/PIP_Cloud.Memo.pdf.pdf
| url-status = dead
}}
#* {{cite journal
| last1 = Yau
Line 736 ⟶ 748:
| last4 = Gibson
| first4 = Garth
| title = Proceedings of the 4th Annual Workshop on Petascale Data Storage
| chapter = DiskReduce: RAID for data-intensive scalable computing
| s2cid = 15194567
| year = 2009
| doi = 10.1145/1713072.1713075
| pages = 6–10
| isbn = 978-1-60558-883-4
}}
#* {{cite book
Line 768 ⟶ 780:
| last3 = Weatherspoon
| first3 = Hakim
| title = Proceedings of the 1st ACM symposium on Cloud computing
| chapter = RACS: A case for cloud storage diversity
| s2cid = 1283873
| year = 2010
| doi = 10.1145/1807128.1807165
Line 818 ⟶ 830:
| isbn = 978-1-4577-1904-2
}}
#* {{cite
| last1 = Qian
| first1 = Haiyang
Line 825 ⟶ 837:
| last3 = T.
| first3 = Trivedi
| title = 12th IFIP/IEEE International Symposium on Integrated Network Management (IM 2011) and Workshops
| chapter = A hierarchical model to evaluate quality of experience of online services hosted by cloud computing
| year = 2011
| doi = 10.1109/INM.2011.5990680
| pages = 105–112
| isbn = 978-1-4244-9219-0
| citeseerx = 10.1.1.190.5148
| s2cid = 15912111
}}
#* {{cite book
Line 854 ⟶ 869:
| chapter = Provable data possession at untrusted stores
| isbn = 978-1-59593-703-2
| url = https://figshare.com/articles/journal_contribution/6469184
}}
#* {{cite book
Line 871 ⟶ 887:
| isbn = 978-1-60558-241-2
| citeseerx = 10.1.1.208.8270
| s2cid = 207170639
}}
#* {{cite book
Line 894 ⟶ 911:
| last2 = S. Kaliski
| first2 = Burton
| title = Proceedings of the 14th ACM conference on Computer and communications security
| chapter = Pors: Proofs of retrievability for large files
| s2cid = 6032317
| year = 2007
| doi = 10.1145/1315245.1315317
Line 916 ⟶ 933:
| chapter = A self-organized, fault-tolerant and scalable replication scheme for cloud storage
| isbn = 978-1-4503-0036-0
| url =http://infoscience.epfl.ch/record/146774
}}
#* {{cite journal
Line 931 ⟶ 949:
| journal=Proceedings of the VLDB Endowment | volume = 2 |issue= 1|doi=10.14778/1687627.1687657
}}
#* {{cite
| last1 = Daniel
| first1 = J. Abadi
Line 976 ⟶ 994:
| chapter = Provable data possession at untrusted stores
| isbn = 978-1-59593-703-2
| url = https://figshare.com/articles/journal_contribution/6469184
}}
# Synchronization
Line 989 ⟶ 1,008:
| doi = 10.1109/CLUSTERWKSP.2010.5613087
| pages = 1–4
| s2cid = 14577793
| chapter = Cloud-based synchronization of distributed file system hierarchies
Line 1,000 ⟶ 1,018:
| s2cid = 16233643
| title = Data Security in the World of Cloud Computing
|
| year = 2009
| doi = 10.1109/MSP.2009.87
Line 1,036 ⟶ 1,054:
| year = 2011
| doi = 10.1109/3PGCIC.2011.37
| s2cid = 13393620
| pages =193–199
| chapter = Suitability of Cloud Computing for Scientific Data Analyzing Applications; an Empirical Study
Line 1,046 ⟶ 1,063:
[[Category:Cloud storage]]
[[Category:Cloud computing]]
| |||