Sunteți pe pagina 1din 69

 |  

    

|   
    




 

   
  

The Evolution Of Data Warehousing

h 
!"#$ %
 
  &  
!  '
&   
 '! (
     
 (!  )
h * ! 
 %+
 
  !

 '   ( 
    ((
!

 , -
%   

 
! (

& )
The Evolution Of Data Warehousing

h ÿ'&%( 
    ' &

 (( !
  !
&

 )
h .
    !

  -

 &   
 )
h /! !( |    ' 
  
  
0
       ! ( 
((
!

  -
%!
&
  
  
(( 
    ! )
Data Warehousing Concepts

h +

! !(|    ' 
&
* 1
 
 
'  2)
h | ! (3
 ( !
( %   (  ! 
 

 
 '  '  4!)
h *
 
!    ( 

|   

    &   
&
 
 )
Data Warehousing Concepts

h /      !!  &!   ' 





*  56   ' 
78)
h |    9
  4!,
 %
  %
,& 
%   ,& 

!!
   
 ((   2 !

  -

(! )
h /     '  

   (
'

  
 
(
 '
!  ! 
 
0
 %(!( % (  
)

h |    
          

!  )
Data Warehousing Concepts

h * !     '  !


'
  ' 

   % 1|   2)
h |   9
 

  '   

(  &'
'
 !    (
)
h /'
 
  ! &
    


&
  
 ! 
''  '
 
'
 )
h |     
 &

! clickstream)
h .
   '   '   !
!-  
   &(  |    )
Oenefits of Data Warehousing

h  

  
&    9
  
 
   !
     ! 

(     '  )
   !   ' :%&:"

 
 & 
 
  &
 )
  *  
 |  ( 
 5*|8
"##;
(  & ,   
&   
5<+*8
  ' 
 !=">)
 #>! (
  !
&&=><+*)
 ÿ ! (
  !
&&";><+*)
 0  !
&  ;><+*)
Oenefits of Data Warehousing

h  (

&&  
 ÿ  
&   ! (
 '
 & !! 
(     '  

&
 !   ! (

&
&    !! (
 
!  )
  (

& &  

 '

!

 , -  !!    ! & 
(&
  &
 % - '    ((

 
  %3 (%!   %  
  )
Oenefits of Data Warehousing

h * !  !


&
+( |!

 
-
 |   

(& (!
&
!( 
!

 , - ! 
 
     
!
 % 4!
 %

!   )
 *
       
(
! ( 
   
 
  (&
  !
 &
'
 
 )
  
  
 

 
 %   
'   ' !( !

  - ( 
   
&% !!  !
  
)
Comparison Of OLTP and Data
Warehousing Systems

h +?/   h |   


 
 ÿ !   )  ÿ 

!   )
  
   
%
% 

  )  
  )
 |  

!)  |  
    
!)
  ÿ!% !% 

!
<(

&(!
)
(!
)
 ÿ
& !
   
 '& !
 
 ()  ()
 
! (     )  . (
! (   )
 / !
 
& )   

& )
 ((
! 
 
 )  4!
 )
 ((   !

 )  ((   
!!

 )
 &       &  
&'  
!
! @( 
   ) 
  )
Comparison Of OLTP and Data
Warehousing Systems

h  +?/     |      &



 !  !

!   
'

 
((  

%     !  )
 +?/    (&
 !  '  )
h  4( 

 
 
(
   
+?/    ! 
!
 %
  %  4!! %! 


(
! 

 
 )
h /%    1! (2

!  
   '  )
Comparison Of OLTP and
Data Warehousing Systems

h +?/        '  %


'
!
 
 )
h |  '   '  ! (3
0
 %
  '  & ! 


0 A=B
  '  & ((  
 !(((
 


AB
Problems Of Data Warehousing

h /'
   (  !

'
&(
 
   '  )
 .  

  !    
)
 ÿ
 ( '
 !   )
 <0
|   ! ()
 * !  ,   )
 |    
 
 )
 ÿ
  ! )
 |  '  
()
 ÿ

 !)
 ? ,| 
 (4! )
  (3

  
 )
Data Warehouse Architecture
   

+( 
  |   <(
%0% ((
! 
 
!" &(  % *
ÿ

    
 C
?     
 

+( 
  |  
!A ?
 
 | 
+? /

|
|
 | 

   
+( 
  |  
!

|  



!
&@ !-(
 
 ,  !! 
+( 
  |  
5+|8
Data Warehouse Architecture

h +( 
 |  9
 !|  |     )
h +( 
 |  9
 (
!  
  +( 
 
|    
)
h ?  9
   ( 
  !
'

3 !
   
  
'  )
Data Warehouse Architecture

h    9


  ( 
  !
h  
   !
 !)
h /  
   
 !   
 (   
  '    )
h  
 
3  &
'     )
h   
   
 
 )
h   
   
 )
h  !-
,(  !
&
  )
Data Warehouse Architecture

h C 
   ( 
  !
'
    
0
 )
 3 ( %
h |
!
0
  (((
  )
h !
3!
 0
 )
h |
| 
 /      
  
    ! )
 D &
  
  &
   
  
3&
)
 +    
%
  
 '  
((      )
Data Warehouse Architecture

h ?
 ÿ
 
| 
 /
     (

 


  
     
'   )
 /
  '  

 

 ( !
0(
 )
 ( 
 ((( !0
 )
    
( ! 
   '
  
 
'  )
Data Warehouse Architecture

h !
&@ !-(| 
  |  '    
   

  ((   !
&
  !-()
 |  
   &
!  !  
!
 ((
! 
-)
h   
         5     8


 
  (!  
'  )
 .  & 
(( 
h 3 !
   
(!  )
h      (! )
h  ( 0   (! )
Data Warehouse Architecture

h  ,. !! /


 .  
 !'
'  
 , 
 )E3 (%
h <(
 0 )
h ((
! 
 &(   )
h 3!
&*  
   5*8 )
h + 
 
!  !
5+? 8 )
h |  

 )
Data Warehouse Data Flows
   

  
+( 
  |   <(
%0% ((
! 
 
!" &(  % *
ÿ

 
 C
Inflow  
?   
 

+'
?
 
 | 
+? /

.('

|
|
 | 

   
+( 
  |  
! |' '
|  



!
&@ !-(
 
 ,  !! 
+( 
  |  
5+|8
Data Warehouse Data Flows

h |   
!   
  
&(
  ' )
 * ')
 .(')
 |' ')
 +')
  ')
Data Warehouse Data Flows

h * '
 /(!   !
'
3 !
 %!
%
 
    !   

  '  )
   !  
   (
+?/ 
  %    ! !((  
  '  )
h /! !
   
&&
 

 
 < !
   
 '0
     
'  
!
%3 (% 
 @ &

 %
  

  )
  
  !  
!
 '

  '

    
'  )
Data Warehouse Data Flows

h .('
 /(!   !
'
 
&   

'     

%( !-
% 



   )
 !
&

  !
'
('
!
h  

   !
%(4!
%4

% 
(
 
   
&
'    ! &
 
  ,  )
h !-
  ! &

  

  
    % !  (   %3
!   %!  %  (
! (   
 %(
& 
     

 )
h |


   (((
 ( 
! 
 
&
 

  !!


)
Data Warehouse Data Flows

h |' '
 /(!   !
'
 !
&
  !-
,(
  
'  )
 !
&
  (   
( 




!
&   ( !'  
 
  

&    
!
& !   
! ((
! 
-)
 /' '  
! (!    
!     '  ! 

'
   % ' @ ' 
 )
 !
&    
 '   ' ,
  
    
'  %' 0
)
Data Warehouse Data Flows

h +'
 /(!   !
'
 -
  
&
  ,  )
 /'- !
&

 
&&
'

!9
h ` '
!
! ! '
 

 ,
  20      )
h | 
%'
!
! ! '
( !
&

&

 
  ,  2'-  
  

  (1(
, ,  !
2
(! )
Data Warehouse Data Flows

h  '
 /(!   !
'
   
    )
  '
(!   &     
5   ' 8)
    
  !
(
   !   
  '  %' 


%'
! 
 

% '    

' !
%
  
%   

)
Data Warehouse Tools And
Technologies

h 3 !
 %
% /  
 
/
  ! 

 (  (! %
 

   
 )
 *    
  
 '

!  
 )
h    )
h |     (
! 
  )
h |
!  
 
 )
Data Warehouse Tools And
Technologies

h |    |
 <0
      '  |
h ? ( !
h ? (!

h |  0 
   
h C( !
h /  !  


h    !  


h D'-  '  
h    

 

h *   

  

h & !0 !
 
)
Data Warehouse Tools And
Technologies

h |     | 
 /       
'  
 3 ! (3


! -)
 / 4
  
 

'  !
& 
 (     
    '  )
 * 

'' 4          


    ' 
 ! (  &(  (( 
 |   

 5|8 +4!   (5+8)
 |  '
+)
 /   

!( 
  |2 +( *  
 
5+*8'
 +2      58)
 / '&
 + 2 D
  & 
     %
!
+2 
.?5.


?  8%F*5F?   * ! 8% +E
5 +4!E !

8%  |2 +*)
 /G' &(   ! (
 %
!
*%+ !%
.
 %ÿ(
 % 
%D<%.% |

 |*)
Data Warehouse Tools And
Technologies

h 

 
     /
 /|     

 
        
! (  ((
'
 - 9
h 

   
  
( ! )
h |  0 
 
 
!!- )
h 
 ( 
    )
h 

   ( ! 
!
 0 ( 
 
 !

 
 )
h 

  '   (&
 !   !-
 
 )
h <(
! 
%  
 


  )
h 



!
        )
h 
  )
h !
&
  !-
,(  )
h * (  
!&'

)
h !
   )
Data Marts

  |  
    (( 
0
    
( 
! (   

  !
 )
Date Warehouse Vs Data Mart

E!    
 E!    
 (
 0
     
!
'
 
(   @
 
 !

 

  
   
+( 
 | 
 


! (3   


   
 &
 &

Why Data Marts?
h /
&   !!      
h /(&
  
    ! !!
&&
'  
  (  
 (   
  !

h /(&
   ( 
!
 
& 
   !! 
h /(&
 (((
 !   
! . 
<0
   !5.<8
 
-+? @|  


h / !

  3!
  -  !   !
% 
%
  
  
  

h   !5! ( |     8
h /
( 
  
 ! '  
( !
    &  ' 

Approaches to building data marts

h +  (( !

 & |    
'
 &
'&  
  
 
 
'  
h + (( !


 ! 
!(   '  '
  



  |   




  
Data Mart Architecture
   

+( 
  |  
!" <(
%0% ((
! 
 
&(  % *
  
?  ÿ

   

 
C
 
+( 
  |   ?
 
 | 
!A
+? /

|
|
 | 

   
+( 
  |  
!
|   
|  



 ,  !! 

+( 
  |  
5+|8 !
&@ !-(  

 
  
Data Mart Issues
h Functionality  
 % ,, !!     %! ( 


  
! (3
     &
! 

h Size 
 
 ( ! 
! 
 ( 
  (! 
'



h Load performance     ( 


        
    !
! 
   



h sers access to multiple data marts 


 ! 
 &
     '
! 

&
'  & (
!    

h Internet/Intranet access ''   !   ! ( &


  
 
3(   ',!  !!    

h Administration       


!  %     

 
 
!  ! (3       &

%    !
 !% (

!
 ( !

Data Warehousing using Oracle9i

h |   
 ((
! 
 0
 (!


(!
!
0  ' ((! (3%
!0

h + !#
 (


 
  

 
 | ! ((
! 

h + 0(!
!
0 % (

! 
0(

 
 % !   !
!!
h (!
  (&
+ !#
 
! 

'
 

Main features of Oracle
h Summary management + !(&
    
 
(

  
 ! ! 
    

h Analytical Functions (&


    !  -
% &
   %(
,
&,(
! ( 
 !

h Oitmapped indexes + !(&


 
!
   
 ((
3  
& !  ! (
 !
0

h Advanced Join methods  ( 



 '
4
%'
!
!  ( !
4
 !
0 ( 


h SQL optimizer +   ('   !)* !  (



   
!
  !! (   4
&0)

h Aesource management + !  !    !


 
 (&
 
!     ! 
  )
Data Warehousing design

h /   ! (  | 


  
 !

 !
0
! |

 

)
h |    
 (( !
    '      )
Dimensionality Modeling

h |


  
! 
!
0 
(  
  
    %


&   ' 
,
( ! !! )

h |  ! !(< 


'
   
!
 )

h &|
! (   ! E !/ %'
 
! (
(
-)
h 

 %
! 
     ! |

 
/  % !!

 
(5  , ! (
8(

-  ' 
'
 
)
Star Schema

TimeID(pk)
Day
E !  week
HHHHHHHHHHH

*|5(-8 ((*|5(-8
((*|5(-8 (
 !*|5(-8 !

+(
!

(
!  !*|5(-8
/(






|   
Dimensionality Modeling cont«

 


!  !  
 ! ! 

 !   

! %  

   
! 

 !  5'
!! 
  
8
Dimensionality Modeling cont«

h  ' -!  & 


  
! '

    
! 
  
 

h   -!  


 ! 
! 
 
3    ' -
!
Part of Snowflake Schema

 !*|5(-8
E !  /(
HHHHHHHHHHH !


*|5-8
((*|5-8
 !*|5-8
+(
!

5(-8

(
! 


ù

|    
Salient Features of DM
h Efficiency /!
 ! 
    ! '  
!
 
!!   & 
  
-('
  0

h Ability to handle changing environments /




  (( !
 0


h Extensibility /
 
3
    
 ' ! %

 % 
 
 -
3



 !  ! ! 

h Ability to model common business situations 


('
 %0  
 
 ! %&
 
   ',     
&

h Predictable query processing &  &  


   ! 

 (


  
! (3%0(!

&(
!  &
 !  0

(  
DM Vs EA modeling

|      ! (  


|    ' <   &
   !
   ! (  
+ 
/ !
  !

  5+?/ 8)
Design Methodology

h 6D
,( 7< (I
 
5"##;8)
h 
|    
 )
h ? ! 
  |    )
ùine-Step Methodology

") 
(!
A) 
 

) * 

 ! 



=) 
 !
) 
(,! ! 
 
 ! 
;) < 


  
$) 
 
    
J) / !-
 '!



#) |!

0(


  0 
Step 1: Choosing the process

h <  4!     )


h   '
   
-9
 |
& 

 

 
  !

( 5  8)
Step 2: Choosing the grain

h |!

3 !'   ! !
(   )
h D- ' 
 ! 
! 
 


 
 ! )
h  %  !  



 


 
  )
Step 3: Identifying and conforming
the dimensions

h !
    

  
    )
h * 

 !! 
'   
   3 ! 5 
 
!    8)
h * 

 

      



conformed)
Step 4: Choosing the facts

h  !   3(  &



(
 
)
h E !    
!  

&)
h 

  ! !    
)
Step 5: Storing pre-calculations in
the fact table

h   (
@      )
h  (,! ! 
  
! ! 
 ' )
h  %   
3  (
  )
Step 6: Aounding out the
dimension tables

h   3 !
(
 


  (
)
h 


&     
  )
Step 7: Choosing the duration of
the database

h ÿ'  !-



 !   )
h /   (    
'
 & 
@
(

 )
h < 
6 '!


 )7
Step 8: Tracking slowly changing
dimensions

h * ( 

  & 
 0
 )
h / %   & 

)
Three types of slowly changing
dimensions

h /("9 ! 



  


&'
 )
h /(A9 ! 

  

!     '

 !! )
h /(9 ! 

  

!       
!   
  '&   
 

   !!
5
 

 8)
Step 9: Deciding the query
priorities and the query modes

h 
!   ! )
h ,   
   
 
 ! )
h  % !
   
! ! 5 !-( %!8)
Assessing the Dimensionality of a
Data Warehouse

h < (I
 5A %8)
h   !& 2 (! )
h A!

 ! !'
 ")
h |
&

!  
 9 !
! %


 
 %3(
 )
Architectural Criteria

h E    !  !

! ' 
  2 
 
 )
h E  !-, % |%
 ,   2  -()
Architectural Criteria

h 3(
!
!  

h   

   !
h |

 
 

h +(    &


h |

   
h ( 
 !
Administration Criteria

h    !*/(  5 (


!-  

! '-'
8)
h ! 6 
7  
'  )
Administration Criteria

h  ! 

! 

h |

 (
! 

h  

  

! 

h  - 

 

h *  
 !
 !
h 
(,

 
 !

h < ,

 
 !

Expression Criteria

h  
!! ( 


 )
h   '       
5 -  
 ((8)
Expression Criteria

h 
(& 


h '!



h <  


h ÿ, ' (( 


h + ,, ! 


h + ,, &



Wash  and Data Warehouses

h ö
3
 2   !
&&   
'  
h   
! ( 
  

,

 % 
,&    
!  
 
  ( !
! 
 )
h (9@@''')! )' )@K! @  !)
MetLife¶s Data Warehouse

h  ' !  !  


( ! 


 ( 
 !)
h E  
 
 
'  &
! 
!  
 )
h (9@@ !
& )! )! @A"@/ÿ@
 @"A@@

)   )
 @
3) 
h <

 0 AA)
 !-& L)"
   !-& L)#5""@A@8
Questions?