16841113 DatawareHousing Concepts

download 16841113 DatawareHousing Concepts

of 20

Transcript of 16841113 DatawareHousing Concepts

  • 7/23/2019 16841113 DatawareHousing Concepts

    1/20

    DATA WAREHOUSINGDATA WAREHOUSING

  • 7/23/2019 16841113 DatawareHousing Concepts

    2/20

    Data Warehousing ConceptsData Warehousing Concepts

    What is a Data Warehouse?What is a Data Warehouse? A Data warehouse is a subect!oriente"# integrate"# ti$e!%ar&ing#A Data warehouse is a subect!oriente"# integrate"# ti$e!%ar&ing#

    non!%o'ati'e co''ection o( "ata in support o( the $anage$ent)snon!%o'ati'e co''ection o( "ata in support o( the $anage$ent)s"ecision *$a+ing process ,OR-"ecision *$a+ing process ,OR-

    Data warehouse is a re'ationa' "atabase (or .uer& an" Ana'&sisData warehouse is a re'ationa' "atabase (or .uer& an" Ana'&sisRather than (or transaction processing/Rather than (or transaction processing/

    Subect!oriente",custo$er#pro"ucts#sa'es#etc-Subect!oriente",custo$er#pro"ucts#sa'es#etc- Non!%o'ati'eNon!%o'ati'e Ti$e!%ar&ingTi$e!%ar&ing Integrate"Integrate"

    (William.Inmon 1993)(William.Inmon 1993)

  • 7/23/2019 16841113 DatawareHousing Concepts

    3/20

    Subect!oriente"Subect!oriente"

    A Data Warehouse is organizedA Data Warehouse is organized

    around major subjects, such asaround major subjects, such as

    customer, supplier, product andcustomer, supplier, product and

    sales. Rather than concentrating onsales. Rather than concentrating on

    day-to-day operations andday-to-day operations and

    transaction processing of antransaction processing of an

    organization.organization.

  • 7/23/2019 16841113 DatawareHousing Concepts

    4/20

    Integrate"Integrate"

    A Data Warehouse is usuallyA Data Warehouse is usually

    constructed by integrating multipleconstructed by integrating multiple

    heterogeneous sources, such asheterogeneous sources, such as

    relational databases, at les, and on-relational databases, at les, and on-line transaction records. data cleaningline transaction records. data cleaning

    and data integration techniues areand data integration techniues are

    applied to ensure consistency in namingapplied to ensure consistency in namingcon!entions, encoding structures,con!entions, encoding structures,

    attribute measures and so on.attribute measures and so on.

  • 7/23/2019 16841113 DatawareHousing Concepts

    5/20

    Ti$e 0ariantTi$e 0ariant

    Data are stored to pro!ideData are stored to pro!ide

    information from a historicalinformation from a historical

    perspecti!e "e#ample$ the past % -&'perspecti!e "e#ample$ the past % -&'

    years(. )!ery *ey structure in theyears(. )!ery *ey structure in the

    data +arehouse contains, eitherdata +arehouse contains, either

    implicitly or e#plicitly an element ofimplicitly or e#plicitly an element of

    timetime

  • 7/23/2019 16841113 DatawareHousing Concepts

    6/20

    Non!0o'ati'eNon!0o'ati'e

    A data +arehouse is al+ays a physicallyA data +arehouse is al+ays a physically

    separate store of data transformed from theseparate store of data transformed from the

    application data found in the operationalapplication data found in the operational

    en!ironment. due to this separation, a dataen!ironment. due to this separation, a data+arehouse does not reuire transaction+arehouse does not reuire transaction

    processing,reco!ery,and concurrencyprocessing,reco!ery,and concurrency

    control mechanisms. it usually reuires onlycontrol mechanisms. it usually reuires only

    t+o operations in data accessing $initialt+o operations in data accessing $initialloading loading of data and access of data.loading loading of data and access of data.

  • 7/23/2019 16841113 DatawareHousing Concepts

    7/20

    Co$parison o( O1T2 s&ste$s an"Co$parison o( O1T2 s&ste$s an"

    "ata warehousing s&ste$s"ata warehousing s&ste$s

    OLTP Systems Data warehousing systems

    Ho'"s Current "ata Ho'"s Historica' "ata

    Stores "etai'e" "ata

    stores "etai'e"#'ight'an" high'&

    su$$ari3e" "ata

    Data is "&na$ic Data is 'arge'& static

    Transaction!"ri%en Ana'&sis "ri%en

    App'ication !oriente" Subect !oriente"

    Supports "a&!to!"a& "ecisions Supports strategic "ecisions

    Ser%es 'arge nu$ber o( c'erica'4operationa'users

    Ser%es re'ati%e'& 'ow nu$ber o( $anageria'users

  • 7/23/2019 16841113 DatawareHousing Concepts

    8/20

    Di$ension 5o"e'ingDi$ension 5o"e'ing

    DimensionDimension$ Dimension is a structure$ Dimension is a structure

    +hich consists of le!els, and+hich consists of le!els, and

    hierarchies is dened on each le!el.hierarchies is dened on each le!el.

    )#ample$)#ample$SEX

    MALE FEMALE

  • 7/23/2019 16841113 DatawareHousing Concepts

    9/20

    Di$ension 5o"e'ingDi$ension 5o"e'ing

    )#ample$)#ample$

    Profession

    Engineer Secretary Teacher

    Chemical Civil Eecutive !unior Elementary "igh School

    Level #

    Level $

    Level %

  • 7/23/2019 16841113 DatawareHousing Concepts

    10/20

    6ACTS6ACTS

    act$ act consists of +hole data +ithact$ act consists of +hole data +ith

    primary *ey, foreign *ey relation shipprimary *ey, foreign *ey relation ship

    +ith dimensions and also consists of+ith dimensions and also consists of

    measures.measures.

    here are hree types of factshere are hree types of facts

    &.ADD/) A01&.ADD/) A01

    2.1)3 ADD/) A012.1)3 ADD/) A01

    4.565 ADD/) A014.565 ADD/) A01

  • 7/23/2019 16841113 DatawareHousing Concepts

    11/20

    6act 'ess 6act 7 Con(or$ "i$6act 'ess 6act 7 Con(or$ "i$

    act less fact is a fact it does notact less fact is a fact it does not

    containcontain

    3easures.3easures.

    A Dimension +hich can share moreA Dimension +hich can share morethan one act is called 0onformthan one act is called 0onform

    DimensionDimension

    0ollection of 1tar 1chemas and0ollection of 1tar 1chemas and1no+a*e1no+a*e

    1chemas is called 7ala#y.1chemas is called 7ala#y.

  • 7/23/2019 16841113 DatawareHousing Concepts

    12/20

    Star sche$a 7 Snow ('a+e Sche$aStar sche$a 7 Snow ('a+e Sche$a

    1tar schema$1tar schema$

    A 0entralized fact table surrounded +ithA 0entralized fact table surrounded +ithdimension tables ha!ing 8rimary, oreign *eydimension tables ha!ing 8rimary, oreign *eyrelation ship bet+een them is called star schema.relation ship bet+een them is called star schema.

    1no+ a*e 1chema$1no+ a*e 1chema$

    A normalized star schema s called 1no+A normalized star schema s called 1no+a*e 1chemaa*e 1chema

    7ala#y$7ala#y$

    0ollection of 1tar schemas and sno+ a*e0ollection of 1tar schemas and sno+ a*e1chema is called 7ala#y.1chema is called 7ala#y.

  • 7/23/2019 16841113 DatawareHousing Concepts

    13/20

    Star Sche$aStar Sche$a

    Fact Ta&le

    Profession'Dim

    Se'Dim Date'Dim

    A((ress Dim

    Se )ey

    Se

    Profession')ey

    Profession'class

    Title

    Level

    (isci*line

    A((ress )ey

    Country

    State

    City

    Date )ey

    Current year

    Current month

    Current wee)

    Con(or$ Di$Profession +ey'''''''''''''''''''''''''''''

    Se )ey

    ''''''''''''''''''''''''''''

    A((ress )ey

    '''''''''''''''''''''''''''''

    Date )ey

    '''''''''''''''''''''''''''''

    ' ''''

    ''''

    '''''

    Measures ,-umeric.

  • 7/23/2019 16841113 DatawareHousing Concepts

    14/20

    Snow 6'a+e Sche$aSnow 6'a+e Sche$a

    Fact Ta&leTime Dim

    City DimLocation DimSu**lier Dim

    /tem Dim

    /tem )ey/tem name

    Ty*e

    Su**lier )ey

    Time )ey/tem )ey

    Location )ey

    ''''''

    ''''''

    ''''''''''''''''''''

    ''''''''

    0u*ees sol(

    1nits sol(

    Time )ey2ear

    3uarter

    Month

    (ay

    Location )ey

    Street

    City )ey

    Su**lier )ey

    Su**lier name

    Su**lier a((ressSu**lier ty*e

    City )ey

    City name

    State

    CountryPin co(e

  • 7/23/2019 16841113 DatawareHousing Concepts

    15/20

    T82ES O6 5A22INGST82ES O6 5A22INGS

    9istory9istory

    1imple pass through "5one(1imple pass through "5one(

    1lo+ly gro+ing target "ull(1lo+ly gro+ing target "ull( 1lo+ly changing dimension1lo+ly changing dimension

    "depends("depends(

  • 7/23/2019 16841113 DatawareHousing Concepts

    16/20

    T&pes o( SCD9ST&pes o( SCD9S

    1lo+ly changing dimension:&1lo+ly changing dimension:&

    1lo+ly changing dimension:21lo+ly changing dimension:2

    &.&. ime stampingime stamping

    2.2. !ersioning!ersioning

    4.4. laginglaging 1lo+ly changing dimension:41lo+ly changing dimension:4

  • 7/23/2019 16841113 DatawareHousing Concepts

    17/20

    S'ow'& changing Di$ension!:S'ow'& changing Di$ension!:

    10D-&$When you does not +ant10D-&$When you does not +ant

    9istory9istory

    use this *ind of mapping "6nly insertuse this *ind of mapping "6nly insert

    elseelse

    ;pdate ta*es place( it inserts the;pdate ta*es place( it inserts the

    ne+ ro+ or ;pdate the e#istingne+ ro+ or ;pdate the e#isting

    dimensions.dimensions.

  • 7/23/2019 16841113 DatawareHousing Concepts

    18/20

    S'ow'& changing Di$ension!;S'ow'& changing Di$ension!;

    10D -2 "ime stamp($10D -2 "ime stamp($ When you +antWhen you +antmaintain full history use this *ind of mapping.maintain full history use this *ind of mapping.

    nserts ne+ and changed dimensions. 0reates annserts ne+ and changed dimensions. 0reates an

    e

  • 7/23/2019 16841113 DatawareHousing Concepts

    19/20

    S'ow'& changing Di$ension!

  • 7/23/2019 16841113 DatawareHousing Concepts

    20/20

    Data Warehouse E=ecutionData Warehouse E=ecution

    ArchitectureArchitecture ArchitectureArchitecture

    DB2/400

    Oracle

    S! Ser"er

    #lat $le%

    ODB&

    'ati"e

    #

    In*ormatica (+!),'I-

    ODB&

    'ati"e

    '&era DataWarehou%e

    Source S%tem% +! DAA WA+O,S+

    0e2

    ortin3(DSS)

    Ser"er%

    ODB&