Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 35 views

    Characterizing The File Hosting Ecosystem: A View From The Edge (IFIP Performance 2011)

    Uploaded by

    AsicsNew
    We present a comprehensive, longitudinal characterization study of the file hosting ecosystem using HTTP traces collected from a large campus network over a one-year period. We performed detailed multi-level analysis of the usage behavior, infrastructure properties, content characteristics, and user-perceived performance of the top five services in terms of traffic volume, namely RapidShare, Megaupload, zSHARE, MediaFire, and Hotfile. We carefully devised methods to identify user clickstreams in the HTTP traces, including the identification of free and premium user instances, as well as the identification of content that is split into multiple pieces and downloaded using multiple transactions. Throughout this characterization, we compare and contrast these services with each other as well as with peer-to-peer file sharing and other media sharing services.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    Characterizing The File Hosting Ecosystem: A View From The Edge (IFIP Performance 2011)

    Uploaded by

    AsicsNew
    35 views18 pages
    We present a comprehensive, longitudinal characterization study of the file hosting ecosystem using HTTP traces collected from a large campus network over a one-year period. We performed detailed multi-level analysis of the usage behavior, infrastructure properties, content characteristics, and user-perceived performance of the top five services in terms of traffic volume, namely RapidShare, Megaupload, zSHARE, MediaFire, and Hotfile. We carefully devised methods to identify user clickstreams in the HTTP traces, including the identification of free and premium user instances, as well as the identification of content that is split into multiple pieces and downloaded using multiple transactions. Throughout this characterization, we compare and contrast these services with each other as well as with peer-to-peer file sharing and other media sharing services.

    Original Title

    Characterizing the file hosting ecosystem: A view from the edge (IFIP Performance 2011)

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    We present a comprehensive, longitudinal characterization study of the file hosting ecosystem using HTTP traces collected from a large campus network over a one-year period. We performed detailed multi-level analysis of the usage behavior, infrastructure properties, content characteristics, and user-perceived performance of the top five services in terms of traffic volume, namely RapidShare, Megaupload, zSHARE, MediaFire, and Hotfile. We carefully devised methods to identify user clickstreams in the HTTP traces, including the identification of free and premium user instances, as well as the identification of content that is split into multiple pieces and downloaded using multiple transactions. Throughout this characterization, we compare and contrast these services with each other as well as with peer-to-peer file sharing and other media sharing services.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    35 views18 pages

    Characterizing The File Hosting Ecosystem: A View From The Edge (IFIP Performance 2011)

    Uploaded by

    AsicsNew
    We present a comprehensive, longitudinal characterization study of the file hosting ecosystem using HTTP traces collected from a large campus network over a one-year period. We performed detailed multi-level analysis of the usage behavior, infrastructure properties, content characteristics, and user-perceived performance of the top five services in terms of traffic volume, namely RapidShare, Megaupload, zSHARE, MediaFire, and Hotfile. We carefully devised methods to identify user clickstreams in the HTTP traces, including the identification of free and premium user instances, as well as the identification of content that is split into multiple pieces and downloaded using multiple transactions. Throughout this characterization, we compare and contrast these services with each other as well as with peer-to-peer file sharing and other media sharing services.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 18

    CharacLerlzlng Lhe llle PosLlng

    LcosysLem A vlew from Lhe Ldge


    An|ket Mahant| Un|vers|ty of Ca|gary Canada
    Carey Wllllamson unlverslLy of Calgary Canada
    nlklas Carlsson Llnkplngs unlverslLeL Sweden
    MarLln ArllLL P Labs uSA
    Anlrban MahanLl nlC1A AusLralla
    lnLroducLlon
    W usergeneraLed conLenL has Lransformed how
    people share and dlssemlnaLe lnformaLlon
    W Web has wlLnessed emergence of flle hosLlng
    servlces
    opular servlces lnclude 8apldShare and Megaupload
    W llle hosLlng servlces offer several advanLages over
    2 flle sharlng and newage conLenL sharlng
    sysLems
    Convenlence hlgher flle avallablllLy lmproved prlvacy
    dlverse conLenL and economlc lncenLlves
    W llle hosLlng servlces offer dlfferenLlaLed servlces
    for free and premlum users
    MoLlvaLlon
    0 3 10
    lpoque 07 (Cermany)
    lpoque 07 (Mld LasL)
    Sandvlne 09 (Clobal)
    lpoque 09 (Clobal)
    Maler eL al lMC 09 (Luropean lS)
    AnLonlades eL al lMC 09 (Creek unlv)
    ||e nost|ng 1raff|c ( of Internet 1raff|c Vo|ume)
    LabovlLz eL alSlCCCMM 10
    2 traff|c (71)
    LabovlLz eL alSlCCCMM 10
    ||e host|ng traff|c
    Growth between May]09 and May]10
    1ype S|te Users () V|s|ts ()
    ||e
    host|ng
    8apldShare 61 79
    Megaupload 128 203
    Soc|a|
    med|a
    ?ou1ube 28 68
    lacebook 43 69
    llle PosLlng LcosysLem
    Cb[ecLlves
    W resenL a comprehenslve longlLudlnal
    characLerlzaLlon sLudy of flle hosLlng servlces
    uaLa collecLed from a campus neLwork over 1year
    perlod
    W erform mulLllevel analysls of usage behavlour
    lnfrasLrucLure conLenL characLerlsLlcs and
    userpercelved performance
    W Analyze ln deLall Lhe 1op5 setvlces by Lrafflc
    volume
    uaLaseLs
    W P11 LransacLlons Lrace collecLed from a large
    campus neLwork wlLh 33k users beLween !an/09
    and uec/09
    W used 8ro's parslng capablllLles Lo summarlze
    P11 requesLresponse palrs ln real Llme
    AppllcaLlon layer (P11 headers) and LransporL layer
    (byLes Lransferred)
    W user ldenLlflable lnformaLlon such as l and
    cookles noL sLored
    W AggregaLed Lhe Lraces uslng Lhe P11 PosL
    header for Lhe 1op3 servlces
    uaLa Analysls (P11 1ransacLlons)
    W uevelop slgnaLures Lo dlsLlngulsh free and
    premlum users by leveraglng cllcksLreams
    W Lxample shows how walL Llme ls calculaLed
    1race Cvervlew
    Character|st|c 1opS Serv|ces
    1oLal P11 1ransacLlons 6 mllllon
    number of days of user acLlvlLy 349 (96 of Lhe year)
    number of flles downloaded 93143
    1oLal slze of downloaded flles 8 18
    llle uploaded 393
    1oLal slze of uploaded flles 393 C8
    remlum flle downloads 40
    lree flle downloads 60
    Average flle slze 106 M8
    Average conLenL slze 176 M8
    Campus usage
    W Conslderable growLh ln usage for mosL
    servlces
    W Ma[orlLy of Megaupload and 8apldShare flles
    downloaded by premlum users
    1lme aLLerns
    W usage ls skewed Lowards Lhe evenlng
    W unlform usage durlng Lhe week due Lo users
    ln Lhe campus resldence
    W Clear lncrease ln flle hosLlng Lrafflc volume
    Lowards Lhe end of Lhe calendar year
    Server LocaLlon
    W Servers locaLed uslng Lhe delaybased based
    geolocaLlon (shorLesL plng)
    W LxcepL for Megaupload all oLher servlces have
    cenLrallzed server locaLlons
    Server lnfrasLrucLure
    Serv|ce nost|ng company or
    carr|er
    Locat|on ]24
    subnets
    nost I
    addresses
    8apldShare CogenL Cerman
    1elecom Clobal
    Crosslng 1A1A Level3
    1ellaSonera
    Cermany 48 8700
    Megaupload CarpaLhla LeaseWeb uS/Canada/
    neLherlands
    27 938
    zSPA8L Choopa uS 2 97
    Medlallre CogenL Llnk8lghL uS 3 837
    PoLflle Lemurla WZ Comm uS 6 178
    ConLenL 1ype
    W lor 8apldShare and PoLflle mosL conLenL was archlve
    flles
    W Megaupload had hlgher proporLlon of vldeo conLenL
    whlle Medlallre had large number of M3 flles
    W MosL zSPA8L conLenL was sLreamlng
    W lurLher analysls of archlve conLenL revealed LhaL Lhey
    were mosLly audlo/vldeo flles
    ConLenL lragmenLaLlon and Slze
    W ConLenL ls hlghly fragmenLed across flle
    hosLlng servlces (3080)
    W ConLenL hosLed on 2 ls an order of
    magnlLude larger Lhan flle hosLlng servlces
    ConLenL Sources
    W P11 8eferer header used Lo caLegorlze
    conLenL sources
    W revalence of a wlde varleLy of sources
    lncludlng porLals blogs and forums
    uownload 8aLes
    W remlum download raLes are an order of
    magnlLude hlgher Lhan free download raLes
    W lree download raLes are hlgher Lhan 2
    W Megaupload and Medlallre offer conslderably
    hlgher download raLes for free users
    llle AvallablllLy
    W Crawled 300k flles from an lndexlng slLe and
    recorded Lhe daLe Lhe flle was uploaded
    W Medlan flle avallablllLy ls 4 monLhs hlgher
    Lhan 2 buL lower Lhan vldeo sharlng slLes
    such as ?ou1ube
    Summary
    W resenLed Lhe largesL characLerlzaLlon sLudy of
    Lhe flle hosLlng ecosysLem
    W PlghllghLed sallenL feaLures of Lhe ecosysLem and
    ldenLlfled slmllarlLles and dlfferences beLween
    servlces
    W 8esulLs lndlcaLe LhaL Lhe flle hosLlng ecosysLem ls
    flourlshlng
    W 8esulLs also suggesL LhaL Lhe economlc model
    based on adverLlsemenL and subscrlpLlon ls
    susLalnable

    You might also like