View in EDS

Dynamic single node failure recovery in distributed storage systems.

Saved in:
Bibliographic Details
Title: Dynamic single node failure recovery in distributed storage systems.
Authors: Itani, M.1 may.itani@lau.edu.lb, Sharafeddine, S.2 sanaa.sharafeddine@lau.edu.lb, Elkabani, I.1,3 islam.kabani@bau.edu.lb
Source: Computer Networks. Feb2017, Vol. 113, p84-93. 10p.
Subjects: Memory maps (Computer science), Bandwidths, Dynamic storage allocation (Computer science), Signal processing, Genetic algorithms
Abstract: With the emergence of many erasure coding techniques that help provide reliability in practical distributed storage systems, we use fractional repetition coding on the given data and optimize the allocation of data blocks on system nodes in a way that minimizes the system repair cost. We selected fractional repetition coding due to its simple repair mechanism that minimizes the repair and disk access bandwidths together with the property of un-coded repair process. To minimize the system repair cost, we formulate our problem using incidence matrices and solve it heuristically using genetic algorithms for all possible cases of single node failures. We then address three practical extensions that respectively account for newly arriving blocks, newly arriving nodes and variable priority files. A re-optimization mechanism for the storage allocation matrix is proposed for the first two extensions that can be easily implemented in real time without the need to redistribute original on-node blocks. The third extension is addressed by implementing variable fractional repetition codes which is shown to achieve significant cost reduction. The contributions of the paper are four fold: (i) generating an optimized block distribution scheme among the nodes of a given data center for fixed and variable size blocks; (ii) optimization of storage allocation under dynamic environments with data block arrivals; (iii) optimization of storage allocation with newly added storage nodes; and (iv) generating an effective block distribution scheme among the nodes by accounting for varying priorities among data blocks. We present a wide range of results for the various proposed algorithms and considered scenarios to quantify the achievable performance gains. [ABSTRACT FROM AUTHOR]
Copyright of Computer Networks is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
FullText Text:
  Availability: 0
Header DbId: egs
DbLabel: Engineering Source
An: 120833844
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Dynamic single node failure recovery in distributed storage systems.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Itani%2C+M%2E%22">Itani, M.</searchLink><relatesTo>1</relatesTo><i> may.itani@lau.edu.lb</i><br /><searchLink fieldCode="AR" term="%22Sharafeddine%2C+S%2E%22">Sharafeddine, S.</searchLink><relatesTo>2</relatesTo><i> sanaa.sharafeddine@lau.edu.lb</i><br /><searchLink fieldCode="AR" term="%22Elkabani%2C+I%2E%22">Elkabani, I.</searchLink><relatesTo>1,3</relatesTo><i> islam.kabani@bau.edu.lb</i>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22Computer+Networks%22">Computer Networks</searchLink>. Feb2017, Vol. 113, p84-93. 10p.
– Name: Subject
  Label: Subjects
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Memory+maps+%28Computer+science%29%22">Memory maps (Computer science)</searchLink><br /><searchLink fieldCode="DE" term="%22Bandwidths%22">Bandwidths</searchLink><br /><searchLink fieldCode="DE" term="%22Dynamic+storage+allocation+%28Computer+science%29%22">Dynamic storage allocation (Computer science)</searchLink><br /><searchLink fieldCode="DE" term="%22Signal+processing%22">Signal processing</searchLink><br /><searchLink fieldCode="DE" term="%22Genetic+algorithms%22">Genetic algorithms</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: With the emergence of many erasure coding techniques that help provide reliability in practical distributed storage systems, we use fractional repetition coding on the given data and optimize the allocation of data blocks on system nodes in a way that minimizes the system repair cost. We selected fractional repetition coding due to its simple repair mechanism that minimizes the repair and disk access bandwidths together with the property of un-coded repair process. To minimize the system repair cost, we formulate our problem using incidence matrices and solve it heuristically using genetic algorithms for all possible cases of single node failures. We then address three practical extensions that respectively account for newly arriving blocks, newly arriving nodes and variable priority files. A re-optimization mechanism for the storage allocation matrix is proposed for the first two extensions that can be easily implemented in real time without the need to redistribute original on-node blocks. The third extension is addressed by implementing variable fractional repetition codes which is shown to achieve significant cost reduction. The contributions of the paper are four fold: (i) generating an optimized block distribution scheme among the nodes of a given data center for fixed and variable size blocks; (ii) optimization of storage allocation under dynamic environments with data block arrivals; (iii) optimization of storage allocation with newly added storage nodes; and (iv) generating an effective block distribution scheme among the nodes by accounting for varying priorities among data blocks. We present a wide range of results for the various proposed algorithms and considered scenarios to quantify the achievable performance gains. [ABSTRACT FROM AUTHOR]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: <i>Copyright of Computer Networks is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=120833844
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1016/j.comnet.2016.12.005
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 10
        StartPage: 84
    Subjects:
      – SubjectFull: Memory maps (Computer science)
        Type: general
      – SubjectFull: Bandwidths
        Type: general
      – SubjectFull: Dynamic storage allocation (Computer science)
        Type: general
      – SubjectFull: Signal processing
        Type: general
      – SubjectFull: Genetic algorithms
        Type: general
    Titles:
      – TitleFull: Dynamic single node failure recovery in distributed storage systems.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Itani, M.
      – PersonEntity:
          Name:
            NameFull: Sharafeddine, S.
      – PersonEntity:
          Name:
            NameFull: Elkabani, I.
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 11
              M: 02
              Text: Feb2017
              Type: published
              Y: 2017
          Identifiers:
            – Type: issn-print
              Value: 13891286
          Numbering:
            – Type: volume
              Value: 113
          Titles:
            – TitleFull: Computer Networks
              Type: main
ResultId 1