BASIC DEFINITIONS

DWH: is a repository of integrated information, specifically structured for

queries and analysis. Data and information are extracted from heterogeneous
sources as they are generated. This makes it much easier and more efficient
to run queries over data that originally came from different sources.
Data Mart: is a collection of subject areas organized for decision support
based on the needs of a given department. x: sales, marketing etc. the data
mart are designed to suit the needs of a department. Data mart is much less
granular than the !are house data
Data Warehouse: is used on an enterprise level, !hile data marts are used on a
business division " department level. Data !arehouses are arranged around the
corporate subject areas found in the corporate data model. Data !arehouses contain
more detail information !hile most data marts contain more summarized or aggregated
data.
OLTP: #nline Transaction $rocessing. This is standard, normalized
database structure. #%T$ is designed for Transactions, !hich means that
inserts, updates and deletes must be fast.
OLAP: #nline &nalytical $rocessing. 'ead(only, historical, aggregated
data.
Fact Table: contain the quantitative measures about the business
$age ) of )*
Dimesio Table: descriptive data about the facts +business,
Co!orme" "imesios: dimension table shared by fact tables... these
tables connect separate star schemas into an enterprise star schema.
Star Schema: is a set of tables comprised of a single, central fact table
surrounded by de(normalized dimensions. -tar schema implement
dimensional data structures !ith de(normalized dimensions
So# Fla$e: is a set of tables comprised of a single, central fact table
surrounded by normalized dimension hierarchies. -no!flake schema
implements dimensional data structures !ith fully normailized dimensions.
Sta%i% Area: it is the !ork place !here ra! data is brought in, cleaned,
combined, archived and exported to one or more data marts. The purpose of
data staging area is to get data ready for loading into a presentation layer.
&ueries: The D./ contains 0 types of queries. There !ill be fixed queries
that are clearly defined and !ell understood, such as regular reports, canned
queries and common aggregations.
There !ill also be ad hoc queries that are unpredictable, both in quantity and
frequency.
A" Hoc &uer': 1s the starting point for any analysis into a database. The
ability to run any query !hen desired and expect a reasonable response that
$age 0 of )*
makes the data !arehouse !orth!hile and makes the design such a
significant challenge.
The end(user access tools are capable of automatically generating the
database query that ans!ers any question posted by the user.
Cae" &ueries: are pre(defined queries. 2anned queries contain prompts
that allo! you to customize the query for your specific needs
(imball )Bottom u*+ ,s- Imo )To* "o#+ a**roaches:
Acc. To Ralph Kimball, when you plan to design analytical
solutions for an enterprise, try building data marts. When
you have 3 or 4 such data marts, you would be having an
enterprise wide data warehouse built up automatically
without time and effort from eclusively spent on building
the !"W#. $ecause the time re%uired for building a data
mart is lesser than for an !"W#.
INMON : try to build an nterprise !ide Data !arehouse first and all the
data marts !ill be the subsets of the D./. &cc. To him, independent data
marts cannot make up an enterprise data !arehouse under any circumstance,
but they !ill remain isolated pieces of information 3stove pieces
$age 4 of )*
INTRODUCTION TO DATA WAREHOUSE
Itro"uctio
Busiesses o! all si.es a" i "i!!eret i"ustries/ as #ell as %o,ermet
a%ecies/ are !i"i% that the' ca reali.e si%i!icat bee!its b'
im*lemeti% a "ata #arehouse- It is %eerall' acce*te" that "ata
#arehousi% *ro,i"es a e0cellet a**roach !or tras!ormi% the ,ast
amouts o! "ata that e0ist i these or%ai.atios ito use!ul a" reliable
i!ormatio !or %etti% as#ers to their 1uestios a" to su**ort the
"ecisio ma$i% *rocess- A "ata #arehouse *ro,i"es the base !or the
*o#er!ul "ata aal'sis techi1ues that are a,ailable to"a' such as "ata
mii% a" multi"imesioal aal'sis/ as #ell as the more tra"itioal
1uer' a" re*orti%- Ma$i% use o! these techi1ues alo% #ith "ata
#arehousi% ca result i easier access to the i!ormatio 'ou ee" !or
more i!orme" "ecisio ma$i%- Ho#e,er/ at the e" o! all the research/
*lai%/ a" architecti%/ 'ou #ill come to reali.e that it all starts #ith
a !irm !ou"atio- Whether 'ou are buil"i% a lar%e cetrali.e" "ata
#arehouse/ oe or more smaller "istribute" "ata #arehouses
sometimes calle" "ata marts+/ or some combiatio o! the t#o/ 'ou #ill
al#a's come to the *oit #here 'ou must "eci"e o ho# the "ata is to be
structure"- This is/ a!ter all/ oe o! the most $e' coce*ts i "ata
#arehousi% a" #hat "i!!eretiates it !rom the more t'*ical
o*eratioal "atabase a" "ecisio su**ort a**licatio buil"i%- That is/
'ou structure the "ata a" buil" a**licatios arou" it rather tha
structuri% a**licatios a" bri%i% "ata to them-
$age 5 of )*
Data #arehouse mo"eli% is a *rocess that *ro"uces abstract "ata
mo"els !or oe or more "atabase com*oets o! the "ata #arehouse- It
is oe *art o! the o,erall "ata #arehouse "e,elo*met *rocess/ #hich is
com*rise" o! other ma2or *rocesses such as "ata #arehouse
architecture/ "esi%/ a" costructio- We cosi"er the "ata #arehouse
mo"eli% *rocess to cosist o! all tas$s relate" to re1uiremets
%atheri%/ aal'sis/ ,ali"atio/ a" mo"eli%- T'*icall' !or "ata
#arehouse "e,elo*met/ these tas$s are "i!!icult to se*arate- The
se*aratio bet#ee mo"eli% a" "esi% is "oe !or *ractical reasos: it
is our itetio to co,er the mo"eli% acti,ities a" techi1ues 1uite
e0tesi,el'-
Data Warehousi%
Data #arehousi% as more tha 2ust a *ro"uct/ or set o! *ro"ucts3it is
a solutio4 It is a i!ormatio e,iromet that is se*arate !rom the
more t'*ical trasactio5oriete" o*eratioal e,iromet- Data
#arehousi% is/ i a" o! itsel!/ a i!ormatio e,iromet that is
e,ol,i% as a critical resource i to"a'6s or%ai.atios- O!te #e thi$
that a "ata #arehouse is a *ro"uct/ or %rou* o! *ro"ucts/ that #e ca
bu' to hel* %et as#ers to our 1uestios a" im*ro,e our "ecisio5
ma$i% ca*abilit'- But/ it is ot so sim*le- A "ata #arehouse ca hel* us
%et as#ers !or better "ecisio ma$i%/ but it is ol' oe *art o! a more
%lobal set o! *rocesses- These are all 1uestios that must be as#ere"
be!ore a "ata #arehouse ca be built- We *re!er to "iscuss the more
$age 6 of )*
%lobal e,iromet/ a" #e re!er to it as "ata #arehousi%- Data
#arehousi% is the "esi% a" im*lemetatio o! *rocesses/ tools/ a"
!acilities to maa%e a" "eli,er com*lete/ timel'/ accurate/ a"
u"ersta"able i!ormatio !or "ecisio ma$i%- It iclu"es all the
acti,ities that ma$e it *ossible !or a or%ai.atio to create/ maa%e/
a" maitai a "ata #arehouse or "ata mart-
Wh' Data Warehousi%7
The coce*t o! "ata #arehousi% has e,ol,e" out o! the ee" !or eas'
access to a structure" store o! 1ualit' "ata that ca be use" !or "ecisio
ma$i%- It is %loball' acce*te" that i!ormatio is a ,er' *o#er!ul asset
that ca *ro,i"e si%i!icat bee!its to a' or%ai.atio a" a
com*etiti,e a",ata%e i the busiess #orl"- Or%ai.atios ha,e ,ast
amouts o! "ata but ha,e !ou" it icreasi%l' "i!!icult to access it a"
ma$e use o! it- This is because it is i ma' "i!!eret !ormats/ e0ists o
ma' "i!!eret *lat!orms/ a" resi"es i ma' "i!!eret !ile a"
"atabase structures "e,elo*e" b' "i!!eret ,e"ors- Thus or%ai.atios
ha,e ha" to #rite a" maitai *erha*s hu"re"s o! *ro%rams that are
use" to e0tract/ *re*are/ a" cosoli"ate "ata !or use b' ma' "i!!eret
a**licatios !or aal'sis a" re*orti%- Also/ "ecisio ma$ers o!te
#at to "i% "ee*er ito the "ata oce iitial !i"i%s are ma"e- This
#oul" t'*icall' re1uire mo"i!icatio o! the e0tract *ro%rams or
"e,elo*met o! e# oes- This *rocess is costl'/ ie!!iciet/ a" ,er'
time cosumi%- Data #arehousi% o!!ers a better a**roach- Data
#arehousi% im*lemets the *rocess to access hetero%eeous "ata
sources8 clea/ !ilter/ a" tras!orm the "ata8 a" store the "ata i a
$age 7 of )*
structure that is eas' to access/ u"ersta"/ a" use- The "ata is the
use" !or 1uer'/ re*orti%/ a" "ata aal'sis- As such/ the access/ use/
techolo%'/ a" *er!ormace re1uiremets are com*letel' "i!!eret
!rom those i a trasactio5oriete" o*eratioal e,iromet- The
,olume o! "ata i "ata #arehousi% ca be ,er' hi%h/ *articularl' #he
cosi"eri% the re1uiremets !or historical "ata aal'sis- Data aal'sis
*ro%rams are o!te re1uire" to sca ,ast amouts o! that "ata/ #hich
coul" result i a e%ati,e im*act o o*eratioal a**licatios/ #hich are
more *er!ormace sesiti,e- There!ore/ there is a re1uiremet to
se*arate the t#o e,iromets to miimi.e co!licts a" "e%ra"atio o!
*er!ormace i the o*eratioal e,iromet-
Short Histor'
The ori%i o! the coce*t o! "ata #arehousi% ca be trace" bac$ to the
earl' 9:;<s/ #he relatioal "atabase maa%emet s'stems emer%e" as
commercial *ro"ucts- The !ou"atio o! the relatioal mo"el #ith its
sim*licit'/ to%ether #ith the 1uer' ca*abilities *ro,i"e" b' the S&L
la%ua%e/ su**orte" the %ro#i% iterest i #hat the #as calle" e"5
user com*uti% or "ecisio su**ort- To su**ort e"5user com*uti%
e,iromets/ "ata #as e0tracte" !rom the or%ai.atios olie
"atabases a" store" i e#l' create" "atabase s'stems "e"icate" to
su**orti% a" hoc e"5user 1ueries a" re*orti% !uctios o! all $i"s-
Oe o! the *rime cocers u"erl'i% the creatio o! these s'stems #as
the *er!ormace im*act o! e"5user com*uti% o the o*eratioal "ata
*rocessi% s'stems- This cocer *rom*te" the re1uiremet to se*arate
$age 8 of )*
e"5user com*uti% s'stems !rom trasactioal *rocessi% s'stems- I
those earl' "a's o! "ata #arehousi%/ the e0tracts o! o*eratioal "ata
#ere usuall' sa*shots or subsets o! the o*eratioal "ata- These
sa*shots #ere loa"e" i a e"5user com*uti% )or "ecisio su**ort+
"atabase s'stem o a re%ular basis/ *erha*s oce a #ee$ or oce *er
moth- Sometimes a limite" umber o! ,ersios o! these sa*shots #ere
e,e accumulate" i the s'stem #hile access #as *ro,i"e" to e" users
e1ui**e" #ith 1uer' a" re*orti% tools- Data mo"eli% !or these
"ecisio su**ort "atabase s'stems #as ot much o! a cocer- Data
mo"els !or these "ecisio su**ort s'stems t'*icall' matche" the "ata
mo"els o! the o*eratioal s'stems because/ a!ter all/ the' #ere e0tracte"
sa*shots a'ho#- Oe o! the !re1uetl' occurri% remo"eli% issues
the #as to =ormali.>?the "ata to elimiate the ast' e!!ects o! "esi%
techi1ues that ha" bee a**lie" o the o*eratioal s'stems to
ma0imi.e their *er!ormace/ to elimiate co"e tables that #ere "i!!icult
to u"ersta"/ alo% #ith other local cleau* acti,ities- But b' a"
lar%e/ the "ecisio su**ort "ata mo"els #ere techical i ature a"
*rimaril' cocere" #ith *ro,i"i% "ata a,ailable i the o*eratioal
a**licatio s'stems to the "ecisio su**ort e,iromet- The role a"
*ur*ose o! "ata #arehouses i the "ata *rocessi% i"ustr' ha,e
e,ol,e" cosi"erabl' sice those earl' "a's a" are still e,ol,i%
ra*i"l'- Com*ari% to"a'6s "ata #arehouses #ith the earl' "a'6s
"ecisio su**ort "atabases shoul" be "oe #ith %reat care- Data
#arehouses shoul" o lo%er be i"eti!ie" #ith "atabase s'stems that
su**ort e"5user 1ueries a" re*orti% !uctios- The' shoul" o
lo%er be cocei,e" as sa*shots o! o*eratioal "ata-
$age 9 of )*
Data #arehouse "atabases shoul" be cosi"ere" as e# sources o!
i!ormatio/ cocei,e" !or use b' the #hole or%ai.atio or !or smaller
commuities o! users a" "ata aal'sts #ithi the or%ai.atio- Sim*l'
ree%ieeri% source "ata mo"els i the tra"itioal #a' #ill o lo%er
satis!' the re1uiremets !or "ata #arehousi%- De,elo*i% "ata
#arehouses re1uires a much more thou%ht!ull' a**lie" set o! mo"eli%
techi1ues a" a much closer #or$i% relatioshi* #ith the busiess
si"e o! the or%ai.atio- Data #arehouses shoul" also be cocei,e" o! as
sources o! e# i!ormatio- This statemet sou"s cotro,ersial at !irst/
because there is %lobal a%reemet that "ata #arehouses are rea"5ol'
"atabase s'stems- The *oit is/ that b' accumulati% a" cosoli"ati%
"ata !rom "i!!eret sources/ a" b' $ee*i% this historical "ata i the
#arehouse/ e# i!ormatio about the busiess/ com*etitors/
customers/ su**liers/ the beha,ior o! the or%ai.atios busiess
*rocesses/ a" so !orth/ ca be u,eile"- The ,alue o! a "ata #arehouse
is o lo%er i bei% able to "o a" hoc 1uer' a" re*orti%- The real
,alue is reali.e" #he someoe %ets to #or$ #ith the "ata i the
#arehouse a" "isco,ers thi%s that ma$e a "i!!erece !or the
or%ai.atio/ #hate,er the ob2ecti,e o! the aal'tical #or$ ma' be- To
achie,e such iteresti% results/ sim*l' ree%ieeri% the source "ata
mo"els #ill ot "o-
$age : of )*
$age )* of )*