Marakas-Ch10

  • Upload
    vu-thao

  • View
    217

  • Download
    0

Embed Size (px)

Citation preview

  • 7/24/2019 Marakas-Ch10

    1/22

    Marakas: Decision Support Systems, 2nd Edition 2003, Prentice-Hall Capter !0 - !

    Chapter 10:

    The Data Warehouse

    Decision Support Systems in the

    21st

    Century, 2nd

    Editionby George M. Marakas

  • 7/24/2019 Marakas-Ch10

    2/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 2

    10-1: Stores, Warehouses and Marts

    ( data )arehouse is a co''ection o* integrateddatabases designed to support a D.

    (n operationa' data store +D- stores data*or a speci*ic app'ication. t *eeds the data

    )arehouse a strea! o* desired ra) data.( data !art is a 'o)er%cost, sca'ed%do)n/ersion o* a data )arehouse, usua''ydesigned to support a s!a'' group o* users

    +rather than the entire *ir!-.The !etadata is in*or!ation that is kept aboutthe )arehouse.

  • 7/24/2019 Marakas-Ch10

    3/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % #

    The Data Warehouse Environment

    The organiations 'egacy syste!s and datastores pro/ide data to the data )arehouse or!art.

    During the trans*er o* data *ro! the /arioussources, c'eansing or trans*or!ation !ayoccur, so the data in the DW is !ore uni*or!.

    i!u'taneous'y, !etadata is recorded.

    ina''y, the DW or !art !ay be used to createone or !ore 3persona'4 )arehouses.

  • 7/24/2019 Marakas-Ch10

    4/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 5

    Organizational Data Flow and Data

    Storage om!onents

  • 7/24/2019 Marakas-Ch10

    5/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 6

    hara"teristi"s o# a Data Warehouse

    Subject oriented7 organied based on useIntegrated7 inconsistencies re!o/ed

    Nonvolatile7 stored in read%on'y *or!at

    Time variant7 data are nor!a''y ti!e seriesSummarized7 in decision%usab'e *or!at

    Large volume7 data sets are 8uite 'arge

    Non normalized7 o*ten redundant

    etadata7 data about data are stored

    Data sources7 co!es *ro! nonintegratedsources

  • 7/24/2019 Marakas-Ch10

    6/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 9

    $ Data Warehouse is Su%&e"t Oriented

  • 7/24/2019 Marakas-Ch10

    7/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 %

    Data in a Data Warehouse are 'ntegrated

  • 7/24/2019 Marakas-Ch10

    8/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % ;

    10-(: The Data Warehouse $r"hite"ture

    The architecture consists o* /ariousinterconnected e'e!ents:!perational and e"ternal database layer7

    the source data *or the DW In#ormation access layer7 the too's the end

    user access to e

  • 7/24/2019 Marakas-Ch10

    9/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % =

    The Data Warehouse $r"hite"ture )"ont*+

    (dditiona' 'ayers are:$rocess management layer7 the schedu'er

    or >ob contro''er%pplication messaging layer7 the

    3!idd'e)are4 that transports in*or!ationaround the *ir!

    $hysical data &arehouse layer7 )here theactua' data used in the D are 'ocated

    Data staging layer7 a'' o* the processesnecessary to se'ect, edit, su!!arie and'oad )arehouse data *ro! the operationa'and e

  • 7/24/2019 Marakas-Ch10

    10/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 10

    om!onents o# the Data Warehouse

    $r"hite"ture

  • 7/24/2019 Marakas-Ch10

    11/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 11

    Data Warehousing T!olog

    The virtual data &arehouse7 the end usersha/e direct access to the data stores, usingtoo's enab'ed at the data access 'ayer

    The central data &arehouse7 a sing'ephysica' database contains a'' o* the data *ora speci*ic *unctiona' area

    The distributed data &arehouse7 theco!ponents are distributed across se/era'physica' databases

  • 7/24/2019 Marakas-Ch10

    12/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 12

    10-: Data .ave Data -- The Metadata

    The na!e suggests so!e high%'e/e'techno'ogica' concept, but it rea''y is *air'ysi!p'e. Metadata is 3data about data4.

    With the e!ergence o* the data )arehouseas a decision support structure, the !etadataare considered as !uch a resource as thebusiness data they describe.

    Metadata are abstractions %% they are high'e/e' data that pro/ide concise descriptions o*'o)er%'e/e' data.

  • 7/24/2019 Marakas-Ch10

    13/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 1#

    The Metadata in $"tion

    The !etadata are essentia' ingredients in thetrans*or!ation o* ra) data into kno)'edge.They are the 3keys4 that a''o) us to hand'e thera) data.

    or e

  • 7/24/2019 Marakas-Ch10

    14/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 15

    The /eed #or onsisten" in the

    Metadata

    The data )arehouse is set up *or the bene*ito* business ana'ysts and e

  • 7/24/2019 Marakas-Ch10

    15/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 16

    10-: 'nterviewing the DataMetadata

    E2tra"tion

    Aegard'ess o* the nature o* a 8uery, certainaspects o* the !etadata are i!portant toa'' decision%!akers. o!e o* these are:

    What tab'es, attributes and keys doesthe DW containB

    Where did each set o* data co!e*ro!B

    What trans*or!ations )ere app'ied)ith c'eansingB

  • 7/24/2019 Marakas-Ch10

    16/22

  • 7/24/2019 Marakas-Ch10

    17/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 1

    Co!ponents o* the Metadata

    Trans#ormation maps7 records that sho))hat trans*or!ations )ere app'ied

    '"traction history7 records that sho) )hat

    data )as ana'yed%lgorithms #or summarization7 !ethodsa/ai'ab'e *or aggregating and su!!ariing

    Data o&nership7 records that sho) origin%ccess patterns7 records that sho) )hatdata are accessed and ho) o*ten

  • 7/24/2019 Marakas-Ch10

    18/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 1;

    Typica' Mapping Metadata

    Trans*or!ation !apping records inc'ude: denti*ication o* origina' source(ttribute con/ersions

    $hysica' characteristic con/ersionsEncodingre*erence tab'e con/ersionsa!ing changes

    ?ey changesa'ues o* de*au't attributesFogic to choose *ro! !u'tip'e sources('gorith!ic changes

  • 7/24/2019 Marakas-Ch10

    19/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 1=

    10%6: !p'e!enting the Data Warehouse

    (ozar assembled a list o# )seven deadly sins* o#data &arehouse implementation+

    1, )I# you build it- they &ill come*7 the DWneeds to be designed to !eet peop'es

    needs2, !mission o# an architectural #rame&or.7

    you need to consider the nu!ber o* users,/o'u!e o* data, update cyc'e, etc.

    /, 0nderestimating the importance o#documenting assumptions7 theassu!ptions and potentia' con*'icts !ustbe inc'uded in the *ra!e)ork

  • 7/24/2019 Marakas-Ch10

    20/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 20

    3e/en Dead'y ins4, continued

    1, ailure to use the right tool7 a DW pro>ectneeds di**erent too's than those used tode/e'op an app'ication

    2, Li#e cycle abuse7 in a DW, the 'i*e cyc'e

    rea''y ne/er ends/, Ignorance about data con#licts7 reso'/ing

    these takes a 'ot !ore e**ort than !ostpeop'e rea'ie

    , ailure to learn #rom mista.es7 since oneDW pro>ect tends to beget another,'earning *ro! the ear'y !istakes )i'' yie'dhigher 8ua'ity 'ater

  • 7/24/2019 Marakas-Ch10

    21/22

    Marakas: Decision upport yste!s, 2nd Edition "200#, $rentice%&a''

    Chapter 10 % 21

    10-3: Data Warehouse Te"hnologies

    o one current'y o**ers an end%to%end DWso'ution. rganiations buy bits and pieces*ro! a nu!ber o* /endors and hope*u''y

    !ake the! )ork together.(, M, o*t)are (G, n*or!ation ui'dersand $'atinu! o**er so'utions that are at 'east*air'y co!prehensi/e.

    The !arket is /ery co!petiti/e. Tab'e 10%9 inthe te

  • 7/24/2019 Marakas-Ch10

    22/22

    Marakas: Decision upport yste!s, 2nd Edition "200# $ ti & ''

    Chapter 10 % 22

    10%: The uture o* Data Warehousing

    (s the DW beco!es a standard part o* anorganiation, there )i'' be e**orts to *ind ne))ays to use the data. This )i'' 'ike'y bring)ith it se/era' ne) cha''enges:

    Aegu'atory constraints !ay 'i!it the abi'ityto co!bine sources o* disparate data.

    These disparate sources are 'ike'y tocontain unstructured data, )hich is hard to

    store.The nternet !akes it possib'e to access

    data *ro! /irtua''y 3any)here4. * course,this >ust increases the disparity.