Description
Many organizations today have adopted business intelligence (BI) as a catalyst to meet specific business needs and to improve organizational effectiveness.
IBIMA Publishing
Communications of the IBIMAhttp://www.ibimapublishing.com/journals/CIBIMA/cibima.html
Vol. 20 !20"# Article I$ %&'%&# pages
$(I: 0.')/20.%&'%&
Copyright © 2011 In Lih Ong, Pei Hwa Siew and Siew Fan Wong. This is an open aess arti!e distri"#ted
#nder the Creati$e Co%%ons &ttri"#tion Liense #nported '.0, whih per%its #nrestrited #se,
distri"#tion, and reprod#tion in any %edi#%, pro$ided that origina! wor( is proper!y ited. Contat
a#thor In Lih Ong )*%ai!+ ongi!1,%ai!2.#tar.ed#.%y
A Five-Layered Business Intelligence
Architecture
In Lih Ong
1
, Pei Hwa Siew
1
and Siew Fan Wong
2
1
Universiti Tunku Abdul Rahman, Selangor, Malaysia
2
Sunway University, Selangor, Malaysia
--------------------------------------------------------------------------------------------------------------
Abstract
.any organi/ations today ha$e adopted "#siness inte!!igene 01I2 as a ata!yst to %eet spei3i
"#siness needs and to i%pro$e organi/ationa! e33eti$eness. &!tho#gh 1I has "eo%e %ore
ro"#st and per$asi$e, so%e organi/ations are sti!! #na"!e to %a4i%i/e the ret#rn on their 1I
in$est%ents. One ontri"#ting reason is the !a( o3 a good g#iding 1I arhitet#re to s#pport
the i%p!e%entation o3 s#h a syste%. Ha$ing a so!id arhitet#re an he!p organi/ations to
"etter ontro! the i%p!e%entation proess as we!! as the operation o3 the entire 1I
en$iron%ent. & re$iew o3 the e4isting !iterat#re shows that a!tho#gh the i%portane o3 a good
1I arhitet#re is non*arg#a"!e, researh in this area is sti!! !a(ing. To 3i!! the gap, this paper
proposes a 3ra%ewor( o3 1I arhitet#re whih onsists o3 3i$e !ayers+ data so#re, )TL, data
wareho#se, end #ser, and %etadata !ayers. These 3i$e !ayers are essentia! to ens#re high data
5#a!ity and s%ooth in3or%ation 3!ow in a 1I syste%.
Keywords+ "#siness inte!!igene, 1I arhitet#re, 3ra%ewor(, 1I syste%
--------------------------------------------------------------------------------------------------------------
Introduction
1#siness inte!!igene 01I2 is 6a"o#t how to
apt#re, aess, #nderstand, ana!y/e and
t#rn one o3 the %ost $a!#a"!e assets o3 an
enterprise * raw data * into ationa"!e
in3or%ation in order to i%pro$e "#siness
per3or%ane7 0&/$ine et a!., 2008, p. 2182.
It pairs data gathering, data storage, and
(now!edge %anage%ent with ana!yti too!s
to pro$ide deision*%a(ers with
o%petiti$e in3or%ation that o3ten at as
di33erentiators in today9s 3iere "#siness
en$iron%ent 0:egash, 200;2. This e4p!ains
why 1I has re%ained as the top tehno!ogy
priority 3or Chie3 In3or%ation O33iers 3or
the past 3i$e years 0, 200?, 200@, 200Aa, 200A"2.
SpeiBia!!y, a 200A I1. 2. &s s#h, OHS does not store
any historia! data.
%arts to s#pport the%. & data %art is a
s#"set o3 the data wareho#se that is #sed
to s#pport ana!ytia! needs o3 a parti#!ar
"#siness 3#ntion or depart%ent
01#(h"inder et a!., 20082. Li(e data
wareho#ses, it ontains historia! data that
an he!p #sers to aess and ana!y/e
di33erent data trends 0=anEan, 200A2.
Howe$er, it an on!y (eep data 3or >0 to A0
days. There3ore, the a%o#nt o3 data stored
in a data %art is %#h !esser than the data
stored in a data wareho#se. There an "e
%any data %arts inside an organi/ation.
Hata wareho#ses and data %arts are "#i!t
"ased on %#!ti*di%ensiona! data %ode!
whih onsists o3 3at and di%ension
ta"!es. Fat ta"!e ontains 5#antitati$e data
a"o#t "#siness entities s#h as sa!es
a%o#nt, 5#antity, and prie. Hi%ension
ta"!e ontains data 0s#h as prod#t,
#sto%er, data, and !oation2 that
desri"es 3ats 0Fi%"a!! et a!., 200@2.
-etadata Layer
.etadata re3ers to data a"o#t data. It
desri"es where data are "eing #sed and
stored, the so#re o3 data, what hanges
ha$e "een %ade to the data, and how one
piee o3 data re!ates to other in3or%ation
0G
Popo$i et a!., 200A2. These app!iations
an "e #sed to s#pport "oth interna! and
e4terna! "#siness proesses 0Ha$enport D
Mar$enpaa, 200@2. &pp!iations that are
e5#ipped with ana!ytia! apa"i!ities a!!ow
#sers to gain insights into i%pro$ing the
per3or%ane o3 "#siness operations. 1y
e%p!oying ana!ytia! app!iations, deision
%a(ers an a!so identi3y and #nderstand
what 3ators dri$e their "#siness $a!#e, and
th#s a"!e to !e$erage opport#nities 3aster
than their o%petitors 0&$anti(#%ar,
200@2.
For a 1I syste% to wor( s%ooth!y, a!! 3i$e
!ayers desri"ed a"o$e ha$e to "e !in(ed
together in a syste%ati %anner. Hata
originating 3ro% interna! and e4terna!
so#res ha$e to "e e4trated, trans3or%ed,
and !oaded into the data wareho#se !ayer.
When the data passes thro#gh the )TL
!ayer, it an 3!ow to "oth diretions, either
to OHS 0and then to data wareho#se2 or to
data wareho#se diret!y. Sine data
wareho#se is de$e!oped 3or the #sage o3
the entire organi/ation, data 3ro% the
wareho#se is sent to data %arts to 3#!3i!!
spei3i operationa! needs. &t end #ser
!ayer, data in the OHS, data wareho#se, and
data %arts an "e aessed "y #sing a
$ariety o3 too!s s#h as 5#ery and reporting
too!s, data $is#a!isation too!s, and
ana!ytia! app!iations. Fina!!y, there is a
entra!i/ed %etadata repository that is
onneted with $ario#s o%ponents s#h
as the )TL !ayer, the data wareho#se !ayer,
and the end #ser !ayer.
te that the data 3!ows a%ong the
o%ponents as proposed in the 3ra%ewor(
3or this paper are %#!ti*diretiona!. S#h
onsideration o$ero%es !i%itations o3 #ni*
diretiona! data 3!ow in %any e4isting 1I
arhitet#res. & %#!ti*diretiona! 3!ow an
enhane 5#ery per3or%ane and i%pro$e
a#ray "ea#se data error at one !ayer
an "e ret#rned to the pre$io#s !ayer 3or
!ari3iation i3 error o#rs. For instane, i3
an error data is 3o#nd in the )TL !ayer, that
parti#!ar piee o3 data an "e sent "a( to
the data so#re !ayer 0i.e., interna! so#res2
3or %odi3iation. :e$erthe!ess, data 3!ow
3ro% e4terna! so#res to )TL !ayer is on!y
#ni*diretiona!. orretion or
adE#st%ent is 3easi"!e to e4terna! so#re
sine it originates 3ro% the o#tside o3 an
organi/ation.
0onclusion
This paper has proposed a 3ra%ewor( o3
3i$e*!ayered 1I arhitet#re with $ario#s
o%ponents. 1I arhitet#re p!ays an
i%portant ro!e in a33eting the s#ess o3 a
1I i%p!e%entation. To ha$e a s%ooth 1I
operation, organi/ations an "enh%ar(
their arhitet#ra! p!an against the
3ra%ewor( proposed here. 1y ha$ing a
good 1I arhitet#re, organi/ations wi!! "e
a"!e to %a4i%i/e the $a!#e 3ro% their 1I
in$est%ents, and there"y %eet their
"#siness re5#ire%ents and i%pro$e
"#siness per3or%ane. Howe$er, at this
point, the 3ra%ewor( proposed in this
paper re%ains onept#a! in nat#re.
Tho#gh it is "#i!t "ased on e4isting
!iterat#re, the 3ra%ewor( sti!! needs to "e
$a!idated #sing rea!*!i3e 1I ases to a33ir%
its #sa"i!ity. F#t#re researh there3ore an
go a!ong this !ine to $a!idate the
3ra%ewor(.
Acknowledgment
The wor( 3or this paper is s#pported "y
3#nding 3ro% F#nda%enta! =esearh &*;H)0*;;H@*1??@*
>'01)'&>??&2Q0Q&rh1I&I.S.pd3.
Ha$enport, T. H. D Mar$enpaa, S. 0200@2.
VStrategi Ise o3 &na!ytis in 0'*8>0%*.
:egash, S. 0200;2. 61#siness Inte!!igene,7
Communications of the Association for
Information /-stems, 1', 1??*1A8.
:e!son, '10.
Strand, .., Wang!er, 1. D O!sson, .. 0200'2.
NInorporating )4terna! Hata into Hata
Wareho#ses+ Charateri/ing and
Categori/ing S#pp!iers and Types o3
)4terna! Hata,N Proeedings o3 the Ath
&%erias Con3erene on In3or%ation
Syste%s, 2;>0*2;>@.
T#r"an, )., Sharda, =., &ronson, M. ). D Fing,
H. 0200@2. 1#siness Inte!!igene+ &
.anageria! &pproah, 6rentice 9all, Ipper
Sadd!e =i$er, :ew Mersey.
Wa!(er, H. .. 0200>2. 6O$er$iew
&rhitet#re 3or )nterprise Hata
Wareho#ses,7 OOn!ineP, O=etrie$ed
Hee%"er 1, 2010P,
http+QQwww.data%g%t.o%Qinde4.phpR%o
d#!eSdo#%entsDM&S-Ho#%ent
.anager-opSdown!oadFi!eDM&S-Fi!e-idS2A
Wang, H. D Je, T. 020102. 6&n )TL Ser$ies
Fra%ewor( 1ased on .etadata,7 2n+
International ;orAshop on Intelligent
/-stems an+ Applications, 1*;.
Wang, M., Chen, T.*M. D Chi#, S.*H. 020082.
6Literat#re =e$iew on Hata Wareho#se
He$e!op%ent,7 I&CIS PaiBi 2008
Con3erene Progra%, A@?*AA;.
Watson, H. M. 0200A2. 6T#toria!+ 1#siness
Inte!!igene W Past, Present, and F#t#re,7
Communications of the Association for
Information /-stems, 28, ;@?*810.
doc_232250056.pdf
Many organizations today have adopted business intelligence (BI) as a catalyst to meet specific business needs and to improve organizational effectiveness.
IBIMA Publishing
Communications of the IBIMAhttp://www.ibimapublishing.com/journals/CIBIMA/cibima.html
Vol. 20 !20"# Article I$ %&'%&# pages
$(I: 0.')/20.%&'%&
Copyright © 2011 In Lih Ong, Pei Hwa Siew and Siew Fan Wong. This is an open aess arti!e distri"#ted
#nder the Creati$e Co%%ons &ttri"#tion Liense #nported '.0, whih per%its #nrestrited #se,
distri"#tion, and reprod#tion in any %edi#%, pro$ided that origina! wor( is proper!y ited. Contat
a#thor In Lih Ong )*%ai!+ ongi!1,%ai!2.#tar.ed#.%y
A Five-Layered Business Intelligence
Architecture
In Lih Ong
1
, Pei Hwa Siew
1
and Siew Fan Wong
2
1
Universiti Tunku Abdul Rahman, Selangor, Malaysia
2
Sunway University, Selangor, Malaysia
--------------------------------------------------------------------------------------------------------------
Abstract
.any organi/ations today ha$e adopted "#siness inte!!igene 01I2 as a ata!yst to %eet spei3i
"#siness needs and to i%pro$e organi/ationa! e33eti$eness. &!tho#gh 1I has "eo%e %ore
ro"#st and per$asi$e, so%e organi/ations are sti!! #na"!e to %a4i%i/e the ret#rn on their 1I
in$est%ents. One ontri"#ting reason is the !a( o3 a good g#iding 1I arhitet#re to s#pport
the i%p!e%entation o3 s#h a syste%. Ha$ing a so!id arhitet#re an he!p organi/ations to
"etter ontro! the i%p!e%entation proess as we!! as the operation o3 the entire 1I
en$iron%ent. & re$iew o3 the e4isting !iterat#re shows that a!tho#gh the i%portane o3 a good
1I arhitet#re is non*arg#a"!e, researh in this area is sti!! !a(ing. To 3i!! the gap, this paper
proposes a 3ra%ewor( o3 1I arhitet#re whih onsists o3 3i$e !ayers+ data so#re, )TL, data
wareho#se, end #ser, and %etadata !ayers. These 3i$e !ayers are essentia! to ens#re high data
5#a!ity and s%ooth in3or%ation 3!ow in a 1I syste%.
Keywords+ "#siness inte!!igene, 1I arhitet#re, 3ra%ewor(, 1I syste%
--------------------------------------------------------------------------------------------------------------
Introduction
1#siness inte!!igene 01I2 is 6a"o#t how to
apt#re, aess, #nderstand, ana!y/e and
t#rn one o3 the %ost $a!#a"!e assets o3 an
enterprise * raw data * into ationa"!e
in3or%ation in order to i%pro$e "#siness
per3or%ane7 0&/$ine et a!., 2008, p. 2182.
It pairs data gathering, data storage, and
(now!edge %anage%ent with ana!yti too!s
to pro$ide deision*%a(ers with
o%petiti$e in3or%ation that o3ten at as
di33erentiators in today9s 3iere "#siness
en$iron%ent 0:egash, 200;2. This e4p!ains
why 1I has re%ained as the top tehno!ogy
priority 3or Chie3 In3or%ation O33iers 3or
the past 3i$e years 0, 200?, 200@, 200Aa, 200A"2.
SpeiBia!!y, a 200A I1. 2. &s s#h, OHS does not store
any historia! data.
%arts to s#pport the%. & data %art is a
s#"set o3 the data wareho#se that is #sed
to s#pport ana!ytia! needs o3 a parti#!ar
"#siness 3#ntion or depart%ent
01#(h"inder et a!., 20082. Li(e data
wareho#ses, it ontains historia! data that
an he!p #sers to aess and ana!y/e
di33erent data trends 0=anEan, 200A2.
Howe$er, it an on!y (eep data 3or >0 to A0
days. There3ore, the a%o#nt o3 data stored
in a data %art is %#h !esser than the data
stored in a data wareho#se. There an "e
%any data %arts inside an organi/ation.
Hata wareho#ses and data %arts are "#i!t
"ased on %#!ti*di%ensiona! data %ode!
whih onsists o3 3at and di%ension
ta"!es. Fat ta"!e ontains 5#antitati$e data
a"o#t "#siness entities s#h as sa!es
a%o#nt, 5#antity, and prie. Hi%ension
ta"!e ontains data 0s#h as prod#t,
#sto%er, data, and !oation2 that
desri"es 3ats 0Fi%"a!! et a!., 200@2.
-etadata Layer
.etadata re3ers to data a"o#t data. It
desri"es where data are "eing #sed and
stored, the so#re o3 data, what hanges
ha$e "een %ade to the data, and how one
piee o3 data re!ates to other in3or%ation
0G
Popo$i et a!., 200A2. These app!iations
an "e #sed to s#pport "oth interna! and
e4terna! "#siness proesses 0Ha$enport D
Mar$enpaa, 200@2. &pp!iations that are
e5#ipped with ana!ytia! apa"i!ities a!!ow
#sers to gain insights into i%pro$ing the
per3or%ane o3 "#siness operations. 1y
e%p!oying ana!ytia! app!iations, deision
%a(ers an a!so identi3y and #nderstand
what 3ators dri$e their "#siness $a!#e, and
th#s a"!e to !e$erage opport#nities 3aster
than their o%petitors 0&$anti(#%ar,
200@2.
For a 1I syste% to wor( s%ooth!y, a!! 3i$e
!ayers desri"ed a"o$e ha$e to "e !in(ed
together in a syste%ati %anner. Hata
originating 3ro% interna! and e4terna!
so#res ha$e to "e e4trated, trans3or%ed,
and !oaded into the data wareho#se !ayer.
When the data passes thro#gh the )TL
!ayer, it an 3!ow to "oth diretions, either
to OHS 0and then to data wareho#se2 or to
data wareho#se diret!y. Sine data
wareho#se is de$e!oped 3or the #sage o3
the entire organi/ation, data 3ro% the
wareho#se is sent to data %arts to 3#!3i!!
spei3i operationa! needs. &t end #ser
!ayer, data in the OHS, data wareho#se, and
data %arts an "e aessed "y #sing a
$ariety o3 too!s s#h as 5#ery and reporting
too!s, data $is#a!isation too!s, and
ana!ytia! app!iations. Fina!!y, there is a
entra!i/ed %etadata repository that is
onneted with $ario#s o%ponents s#h
as the )TL !ayer, the data wareho#se !ayer,
and the end #ser !ayer.
te that the data 3!ows a%ong theo%ponents as proposed in the 3ra%ewor(
3or this paper are %#!ti*diretiona!. S#h
onsideration o$ero%es !i%itations o3 #ni*
diretiona! data 3!ow in %any e4isting 1I
arhitet#res. & %#!ti*diretiona! 3!ow an
enhane 5#ery per3or%ane and i%pro$e
a#ray "ea#se data error at one !ayer
an "e ret#rned to the pre$io#s !ayer 3or
!ari3iation i3 error o#rs. For instane, i3
an error data is 3o#nd in the )TL !ayer, that
parti#!ar piee o3 data an "e sent "a( to
the data so#re !ayer 0i.e., interna! so#res2
3or %odi3iation. :e$erthe!ess, data 3!ow
3ro% e4terna! so#res to )TL !ayer is on!y
#ni*diretiona!.
orretion oradE#st%ent is 3easi"!e to e4terna! so#re
sine it originates 3ro% the o#tside o3 an
organi/ation.
0onclusion
This paper has proposed a 3ra%ewor( o3
3i$e*!ayered 1I arhitet#re with $ario#s
o%ponents. 1I arhitet#re p!ays an
i%portant ro!e in a33eting the s#ess o3 a
1I i%p!e%entation. To ha$e a s%ooth 1I
operation, organi/ations an "enh%ar(
their arhitet#ra! p!an against the
3ra%ewor( proposed here. 1y ha$ing a
good 1I arhitet#re, organi/ations wi!! "e
a"!e to %a4i%i/e the $a!#e 3ro% their 1I
in$est%ents, and there"y %eet their
"#siness re5#ire%ents and i%pro$e
"#siness per3or%ane. Howe$er, at this
point, the 3ra%ewor( proposed in this
paper re%ains onept#a! in nat#re.
Tho#gh it is "#i!t "ased on e4isting
!iterat#re, the 3ra%ewor( sti!! needs to "e
$a!idated #sing rea!*!i3e 1I ases to a33ir%
its #sa"i!ity. F#t#re researh there3ore an
go a!ong this !ine to $a!idate the
3ra%ewor(.
Acknowledgment
The wor( 3or this paper is s#pported "y
3#nding 3ro% F#nda%enta! =esearh &*;H)0*;;H@*1??@*
>'01)'&>??&2Q0Q&rh1I&I.S.pd3.
Ha$enport, T. H. D Mar$enpaa, S. 0200@2.
VStrategi Ise o3 &na!ytis in 0'*8>0%*.
:egash, S. 0200;2. 61#siness Inte!!igene,7
Communications of the Association for
Information /-stems, 1', 1??*1A8.
:e!son, '10.
Strand, .., Wang!er, 1. D O!sson, .. 0200'2.
NInorporating )4terna! Hata into Hata
Wareho#ses+ Charateri/ing and
Categori/ing S#pp!iers and Types o3
)4terna! Hata,N Proeedings o3 the Ath
&%erias Con3erene on In3or%ation
Syste%s, 2;>0*2;>@.
T#r"an, )., Sharda, =., &ronson, M. ). D Fing,
H. 0200@2. 1#siness Inte!!igene+ &
.anageria! &pproah, 6rentice 9all, Ipper
Sadd!e =i$er, :ew Mersey.
Wa!(er, H. .. 0200>2. 6O$er$iew
&rhitet#re 3or )nterprise Hata
Wareho#ses,7 OOn!ineP, O=etrie$ed
Hee%"er 1, 2010P,
http+QQwww.data%g%t.o%Qinde4.phpR%o
d#!eSdo#%entsDM&S-Ho#%ent
.anager-opSdown!oadFi!eDM&S-Fi!e-idS2A
Wang, H. D Je, T. 020102. 6&n )TL Ser$ies
Fra%ewor( 1ased on .etadata,7 2n+
International ;orAshop on Intelligent
/-stems an+ Applications, 1*;.
Wang, M., Chen, T.*M. D Chi#, S.*H. 020082.
6Literat#re =e$iew on Hata Wareho#se
He$e!op%ent,7 I&CIS PaiBi 2008
Con3erene Progra%, A@?*AA;.
Watson, H. M. 0200A2. 6T#toria!+ 1#siness
Inte!!igene W Past, Present, and F#t#re,7
Communications of the Association for
Information /-stems, 28, ;@?*810.
doc_232250056.pdf