I

T12FinalLect
ureNot
es
Ref
erences: Shel
eyandRosenbl
att,
System Anal
ysi
sandDesi 9thEdi
gn( ti
on)
www.syst
emanal
ysanddesi
gn.com

Busi
nessRequi
rement
s

To be successful
,an inf
ormati
on syst
em mustsat i
sfybusiness requirement s,st aywi t
hin budget
,be
completedontime,andmosti mport
antofallbemanagedef fect
ively
.Whenapr ojectdevel
opspr obl
ems,
thereasonstypi
call
yinv
olvebusi
ness,budget,orscheduleissues.Inaddi t
iontopl anningandmanagi ngthe
proj
ect,apr
ojectmanagermustbeabletorecognizeproblemsanddeal wit
ht hem effectiv
ely.

BusinessIssues
Themaj orobjecti
veofeverysystem istopr ovideasol uti
ontoabusi nessprobl
em oroppor tunit
y.Ifthe
syst
em doesnotdot hi
s,t
henitisaf ail
ure regardl
essofposi ti
vereact
ionfr
om users,acceptablebudget
perf
ormance,ort i
mel
ydeliv
ery.Whent hei nf
ormati
onsy stem doesnotmeetbusinessrequir
ement s,causes
mighti udeuni
ncl dent
ifi
edoruncl
earrequirements,i
nadequatelydef
inedscope,
impreci
setargets,
shor t
cutsor
sl
oppy work dur
ing syst
ems anal
ysis,poordesi gn choi
ces,i
nsuff
ici
enttest
ing orinadequatetesti
ng
pr
ocedures,
andlackofchangecont
rolprocedur
es.Syst
emsal sof
ail
becauseofchangesi
nt heorgani
zat
ion
s
cul
ture,f
undi
ng,orobj
ect
ives.Asy
stem thatfal
lsshor
tofbusi
nessneedsal
sopr
oducespr
obl
emsf
oruser
s
andreducesemploy
eemoraleandproduct
ivi
ty.

Project
swithoutcl
earscopedefi
nit
ionsareri
sky,
becausetheyt
endt
oexpandgradual
ly
,withoutspeci
fi
c
authori
zat
ion,i
napr ocesscal
l
edprojectcr
eep.However,
evenwhenapr
ojecti
sclear
lydescri
bed,i
tmustbe
managedconst ant
ly.

BudgetIssues
Costoverrunsty pi
call
yresultf r
om oneormor eofthefol
lowing:
Unreal
isti
cest i
mat esthatar etooopt imisti
corbasedoni ncompl eteinf
ormat i
on
Fai
luretodev elopanaccur at eforecastthatconsider
sallcostsov ertheli
feoftheproj
ect
Poormoni tor
ingofpr ogressandsl owr esponsetoearlywarningsi gnsofproblems
Scheduledelay sduetof actorst hatwerenotf or
eseen
Humanr esour ceissues,includingturnover,i
nadequatetrai
ning,andmot ivat
ion

ScheduleI
ssues
Problemswithti
metablesandpr oj
ectmilest
onescani ndi
cateafai
lur
etor ecognizetaskdependenci
es,
confusi
onbetweeneffortandpr ogr
ess,poormonitor
ingandcontr
olmethods,personal
i
tyconfl
ict
samong
team members,orturnoverofpr oj
ectper
sonnel.Thefail
ureofanITproj
ectalsocanbecausedbypoor
proj
ectmanagementtechniques.

ManagingSy st
emsPr ojects
Ift
hepr oj
ectmanagerf ai
l
st opl
an,staff
,or ganize,supervi
se,communi cat
e,motiv
ate,eval
uate,direct
,and
contr
olproperl
y,thent
hepr oj
ecti
scertai
nt ofail
.Ev enwhenfactorsout
sidehisorhercontr
olcontri
but et
ot he
fai
lur
e,the projectmanageri sresponsibleforr ecogni
zi
ng the ear
lywar ni
ng si
gns and handling them
eff
ecti
vely.

Projectmanagementisachal
lengingtask.Proj
ectmanager smustbeal er
t,techni
cal
lycompetent
,andhighly
resourcef
ul.Theyal
somustbegoodcommuni cat
orswithstronghumanr esourceskil
ls.Aproj
ectmanager
canbepr oudwhenheorshehandl esasuccessf ulprojectthathel
pst hecompanyachi eveit
sbusiness
objecti
ves,suchast
heAppleproductlaunch.Unfor
tunat
ely,pr
oject
scananddogetder ail
edforawidev
ariety
ofreasons.Whenpr obl
emsoccur,thepr oj
ectmanagersabili
tytohandlethesi t
uat
ionbecomesthecriti
cal
factor.

Whenapr ojectmanagerf ir
strecogni zest hatapr ojectisint rouble,whatopt ionsar eav ai
labl
e?Al t
ernativ
es
cani ncl
udet ri
mmi ngthepr ojectrequirement s,addingtothepr oj
ectr esources,del ayi
ngt hepr ojectdeadl i
ne,
andi mprov ingmanagementcont rolsandpr ocedur es.Somet i
mes,whenapr ojectexper iencesdel aysorcost
overruns,thesy st
em st i
l
lcanbedel i
veredont i
meandwi thinbudgeti fsev erallesscr it
icalrequirement sar e
tri
mmed.Thesy st
em canbedel i
veredt osat isfythemostnecessar yr equirement s,andaddi ti
onalf eaturescan
beaddedl aterasapar tofamai ntenanceorenhancementpr oject.Ifapr oj
ecti sint roublebecauseofal ackof
resourcesoror ganizat
ionalsuppor t,managementmi ghtbewi lli
ngt ogi vethepr ojectmor ecommi tmentand
higherpriority.Forexampl e,managementmi ghtagr eetoaddmor epeopl et oapr ojectt hatisbehi ndschedul e.
Addingst aff,howev er,wil
lreducet hepr ojectscompl et
iont i
meonl yiftheaddi t
ionalpeopl ecanbei ntegrat
ed
effecti
vel
yi ntot hedev el
opmentt eam.I fteam member slackexper i
encewi thcer t
ai naspect soft her equir
ed
technology ,tempor ar
yhel pmi ghtbeobt ainedf rom ITconsul tantsorpar t-timest af f
.Addi ngst affcanmean

1
traini
ngandor i
entingt henewpeople,however.Insomesi tuati
ons,addingmor epeopl
etoapr ojectactual
l
y
mi ghtincreaset hetimenecessaryt ocompletethepr oj
ectbecauseofapr i
nci
plecal
ledBrooksLaw.This
i
nt eresti
ngconceptwasst atedbyFreder
ickBrooks,Jr.
,anIBM engineer,
whoobser vedt
hataddingmanpower
toal at
esof t
war eprojectonlymakesitlat
er.Brooksreachedt hi
sconclusi
onwhenhesawt hatnewwor ker
s
onapr oj
ectf i
rsthadt obeeducat edandi nst
ructedbyexi sti
ngempl oyeeswhoseownpr oducti
vit
ywas
reducedaccor dingly.

Sy
stem Anal
ysi
s

Syst
emsanal ysi
sisthesecondoff ivephasesi nthesy stemsdev el
opmentlifecycl
e.Inthepreviousphase,
syst
emspl anni
ng,y
ouconduct edapreli
mi naryinvest
igationt odeterminet
hepr oj
ect
sfeasi
bil
i
ty.Nowy ouwil
l
userequi
rementsmodeling,dataandprocessmodel i
ng, andobj ectmodelingtechni
questorepresentthenew
syst
em.Youal sowil
lconsidervari
ousdev elopmentst r
at egiesforthenewsy stem,andplanforthet r
ansit
ion
tosyst
emsdesi gnt
asks.Thedel i
ver
ablefort hi
sphasei st hesystem requi
rementsdocument.

Theoveral
lobject
iveofthesy
stemsanal
ysi
sphaseistounderst
andtheproposedpr
oject
,ensur
ethati
twi
l
l
suppor
tbusinessrequi
rements,
andbui
ldasol
idfoundat
ionforsyst
em devel
opment
.

Thesy
stemsanal
ysi
sphaseincl
udesthefourmainacti
vi
ti
esshowninFi
gure4-
2:r
equir
ement smodel
i
ng,
dat
a
andprocessmodeli
ng,obj
ectmodel
ing,andconsi
der
ati
onofdevel
opmentstr
ategies.

Alt
hought hewat erf
allmodelshowssequent i
alSDLC phases,iti
snotuncommonf orseveralphases(or
cer
tai
nt askswi thinaphase)tointer
actduringthedevelopmentprocess,justast heywoul dinanadaptive
model.Forexampl e,thi
soccurswhenevernew fact
sarel ear
nedorsy st
em r equi
rementschangedur ingthe
modeli
ng pr ocess.Fi gur
e 4-
2 showst ypi
calinter
act
ion among the three model i
ng tasks:requi
rements
modeli
ng, dataandpr ocessmodeli
ng,andobjectmodel
ing.

REQUI REMENTSMODELI NG
Requi rementModel li
ng involves fact-f
indi
ng t o descr i
be the cur r
entsy st
em and i denti
ficat
ion oft he
requi r
ement sforthenewsy st
em, suchasout puts,i
nput s,pr
ocesses, performance, andsecurit
y .Outputsrefer
toel ectroni
corpr i
ntedinformati
onpr oducedbyt hesy stem.Input srefert onecessar ydatat hatenterst he
sy stem,eithermanual lyorinanaut omat edmanner .Processesr efertot helogicalrulesthatar eappliedt o
transformt hedat aintomeani ngf
ulinformat i
on.Perfor mancer ef
erst osy stem characteri
sti
cssuchasspeed,
vol ume,capacity,avail
abi
lit
y,andreliabil
it
y.Securit
yr eferstohardwar e,software,andpr oceduralcontrol
st hat
saf eguardandpr otectthesystem andi tsdatafrom internalorexternalthreats.

DATAANDPROCESSMODELI NG
DataandProcessModeli
ng,y
ouwil
lcont
inuethemodeli
ngprocessbylear
ninghowtorepr
esentgraphi
call
y
syst
em dat
aandprocessesusi
ngt
radi
ti
onalst
ruct
uredanal
ysi
stechni
ques.Str
uct
uredanal
ysi
sidenti
fi
esthe

2
dat
afl
owi
ngi
ntoapr
ocess,
thebusi
nessr
ulest
hatt
ransf
ormt
hedat
a,andt
her
esul
ti
ngout
putdat
afl
ow.

OBJECTMODELI NGisanotherpopularmodeli
ngtechnique.Whil
estructuredanalysi
streatsprocessesand
dataasseparatecomponents,obj
ect-
ori
entedanalysi
s(O-O)combinesdat aandt heprocessesthatactonthe
datai
ntothi
ngscalledobj
ects.Theseobject
srepresentactualpeopl
e,things,t
ransacti
ons,andev entst
hat
aff
ectthesystem.Duri
ngthesy st
em

DEVELOPMENTSTRATEGI ES
Dev elopmentSt rategies,y ouwi l
lconsi
dervari
ousdev elopmentopt i
onsandpr epareforthetransiti
ont othe
systemsdesi gnphaseoft heSDLC.Youwi l
llear
naboutsof twaret r
ends,acquisi
ti
onanddev elopment
al
ter nat i
ves,outsour cing,andf or
mallydocument i
ngrequirementsforthenewsy stem.Thedel i
verable,orend
product ,ofthesy stemsanal ysi
sphaseisasy stem r
equirementsdocument ,whichisanov er
alldesignforthe
newsy stem.Inaddi t
ion, eachacti
vit
ywithi
nthesy st
emsanal ysisphasehasanendpr oductandoneormor e
milest ones.Asy oul ear ned,proj
ectmanagersusev arioustoolsandt echni
quest ocoordi
natepeopl e,tasks,
ti
met ables,andbudget s.

SystemsAnal ysi
sSkill
s
Youwi llneedst r
onganal yti
calandi nterper
sonalskil
lstobuildanaccur at
emodeloft henew system.
Analyt
icalskill
senabl eyout oidentif
yapr obl
em,evaluatethekeyel ements,anddev elopausefulsol
ution.
Int
erpersonalski l
ls are especi
all
yv aluableto a systems analystwho mustwor k wit
h peopl
e atal l
organi
zationallevel
s,balanceconf l
icti
ngneedsofuser s,andcommuni cateeffect
ively
.Becauseinfor
mat i
on
systemsaf fectpeoplethr
oughoutt hecompany ,y
oushoul dconsi
derteam-ori
entedstrategi
esasyoubegint he
systemsanal ysisphase.

Team-BasedTechni ques:JAD, RAD, andAgi leMet hods

TheITdepart
ment sgoali stodeliv
erthebestpossi bleinf ormat i
onsy stem, att helowestpossiblecost,i
nt he
short
estpossiblet i
me.To achi evet he bestr esults,sy stem dev elopersv iew user s as partnersint he
devel
opmentpr ocess.Gr eat
eruseri nvolvementusual lyr esultsinbet tercommuni cation,fast
erdev el
opment
ti
mes,andmor esat isfi
edusers.Thet raditi
onalmodelf orsy st
emsdev elopmentwasanI Tdepar tmentthat
usedstr
uctur
edanal y si
sandconsul t
eduser sonlywhent heirinputorappr ov alwasneeded.Al thoughtheI T
staf
fsti
l
lhasacent r
alr ol
e,andstructuredanal ysisremai nsapopul armet hodofsy st
emsdev elopment,most
ITmanagersi
nv it
esy stem user
stopar ticipateactivelyinv ar i
ousdev elopmentt asks.

Apopul arexampl ei sj ointapplicati

ondev el
opment( JAD),whi chisauser -
or i
entedtechni
quef orfact-f
inding
and requirement smodel ing.Because iti snotl i
nked to a speci f
icdev elopmentmet hodology,sy stems
developersuseJADwhenev ergroupinputandi nteracti
onar edesired.Anotherpopularuser-or
ient
edmet hod
i
sr apidappl i
cati
ondev elopment( RAD).RADr esembl esacondensedv er
sionoft heentir
eSDLC,wi thuser s
i
nvolved ev ery step oft he way .Whi l
e JAD t ypical
lyf ocuses only on f act-
fi
nding and requirement s
determination,RADpr ov i
desaf ast-
tr
ackappr oacht oaf ullspectr
um ofsy stem developmenttasks,including
pl
anni ng,desi gn,const ruction,and impl ementati
on.Fi nall
y,agilemet hodsr epr
esentar ecentt r
end t hat
str
essesi ntensei nteract i
onbet weensy stem developersanduser s.

Sy
stem Anal
ysi
sPhase
Thesy
stemsanalysisphaseincl
udesthreeacti
vi
ti
es:
requi
rementsmodeling,
dataandprocessmodel i
ng,and
consi
derat
ionofdevelopmentstrat
egies.

Themainobjecti
veist ounder
standt
hepr oposedproj
ect
,ensur
ethatitwil
lsuppor
tbusi
nessr equirements,
andbuil
dasol idfoundati
onforthesyst
emsdesi gnphase.Dur
ingrequi
rementsmodeli
ng,youi denti
fythe
busi
ness-
rel
ated requi
rementsf orthe new infor
mati
on syst
em,i ncl
udi
ng output
s,inputs,pr ocesses,
perf
ormance,andcontr
ols.

Youconsiderscal
abil
it
ytoensur
ethatthesy
stem cansupportfut
uregrowt
handexpansion.Youal
soesti
mate
tot
alcostofowner shi
p(TCO)toidenti
fyal
lcost
s,incl
udingindi
rectcost
s.Popul
arteam-basedappr
oaches
i
ncludeJAD,RAD,andagilemethods.

Jointappli
cat
iondevel
opment(JAD)isapopul ar
,team-basedapproachtofact
-f
indi
ngandrequir
ement
s
modeling.JAD i
nvol
vesaninter
acti
vegroupofuser s,managers,andITprof
essi
onalswhopart
ici
pat
ein
requi
rementsmodeli
nganddevel
opagreatercommitmenttotheproj
ectandt
othei
rcommongoals.

Rapid appl
i
cat
ion devel
opment(RAD)i sat eam-
based t
echni
que t
hatspeeds up i
nfor
mati
on sy
stems
devel
opmentandpr oducesafunct
ioni
nginf
ormati
onsyst
em.RADi sacompletemethodol
ogy
,wit
haf our
-

3
phasel
i
fecy
clet
hatpar
all
elst
het
radi
ti
onal
SDLCphases.

Agilemethodsatt
emptt
odevel
opasy st
em incr
ement
all
y,bybui
ldi
ngaseri
esofprotot
ypesandconst
antl
y
adjusti
ngthem t
ouserr
equi
rement
s.Systemsanal
yst
susev ari
oustool
sandtechni
questomodelsystem
requir
ements.

Unifi
edModel i
ngLanguage(UML)i sawi del
yusedmet hodofvi
suali
zi
nganddocument
ingsoft
war edesi
gn
throught heeyesofthebusinessuser.UMLt ool
si ncl
udeusecasedi agr
amsandsequencediagramsto
representact
ors,
thei
rrol
es,
andt hesequenceoft
ransacti
onst
hatoccurs.

Af uncti
onaldecomposi t
iondi agr am ( FDD)i samodelofbusi nessf unctionsandpr ocesses.ACASEt oolcan
generateasetofdat af low di agr amsdi rectl
yf rom aFDD.Thef act-
fi
ndingpr ocessi ncludesinter viewing,
documentr eview,obser vation,quest ionnaires,sampl ing,andr esearch.Successf uli nterv
iewingrequi resgood
planningandst rongi nterper sonalandcommuni cationski lls.Thesy stemsanal ystmustdeci deont hepeopl e
tointervi
ew,seti ntervi
ewobj ect i
ves,andpr eparef or,conduct ,andanal y
zei ntervi
ews.Theanal ystal somi ght
fi
ndi thelpfultouseoneormor esof t
war etoolsdur ingf act -
finding.Sy stemsanal ystsshoul dcarefullyr ecord
anddocumentf actualinfor mat i
onasi ti
scol l
ected,andv arioussof twaretool scanhel pananal ystv isualize
anddescr i
beani nformat ionsy stem.Thechapt erconcl udedwi thapr ev i
ew ofl ogi calmodeling.Dat aand
processmodel ingi sast ruct ur edanal y si
sappr oacht hatv iewst hesy stem int ermsofdat aandt hepr ocesses
thatactont hatdat a.Obj ectmodel ingi sanappr oacht hatv iewst hesy stem i nt er
msofdat aandt he
processest hatactont hatdat a.

DATAANDPROCESSMODELI
NGTOOLS

Syst
emsanal ystsusemanygr aphicaltechniquestodescr i
beaninf
ormati
onsystem.Onepopul
armethodisto
drawasetofdat afl
owdiagrams.Adat af l
owdi agram (DFD)usesvari
oussymbolstoshowhowt hesyst
em
tr
ansfor
msi nputdataintousefuli
nf ormati
on.Ot hergraphicalt
ool
sincl
udeobjectmodel
s,(
ObjectModeli
ng)
,
andenti
ty-
rel
ationshi
pdiagr
ams, (Dat aDesign).

DATAFLOW DI AGRAMS
Adataf l
owdi agram (DFD)showshowdat amov est
hr oughaninformati
onsystem butdoesnotshowprogr
am
l
ogicorpr ocessi
ngst eps.AsetofDFDspr ovi
desal ogicalmodelthatshowswhatt hesyst
em does,
nothowit
doesit.Thatdistinct
ionisimportantbecausefocusingoni mplementati
onissuesatthispoi
ntwouldrest
ri
ct
yoursearchforthemostef fect
ivesyst
em design.

Adat afl
owdi
agram (DFD)i
satoolt
hatdepi
ctst
hef
lowofdat
athroughasystem andt
heworkorprocessi
ng
perfor
medbythatsystem.Dat
afl
owdiagramsar
eal
socal
ledbubblechar
ts,
transf
ormat
iongr
aph,and
processmodel
.

Di
agr
ammat
icr
epr
esent
ati
onofdat
ahel
psi
ntwoway
s:

1.I
thelpstheanaly
ststoanalyzet
hesy stem andtomodel sy
stem component
s.
2.Ananaly
stcancommuni cat
ebetterwiththeuser.I
tiseasi
erfort
heusertounder
standandhowt
heanal
yst
hasint
erpr
etedhis/herpr
oblem af
terlookingatDFDs.

DFDSymbolsDFDsusefourbasi
csymbol
sthatrepr
esentprocesses,
dataf
lows,dat
ast
ores,
andent
it
ies.
Sev
eral
dif
ferentv
ersi
onsofDFDsymbol
sexist
,buttheyal
lservethesamepurpose.

PROCESSSYMBOL
Apr ocessrecei
vesinputdat
aandproducesout
putthathasadi
ff
erentcont
ent,for
m,orbot
h.Fori
nstance,t
he
pr
ocessf orcalcul
ati
ngpayusestwoinputs(
payrat
eandhoursworked)toproduceoneout
put(t
otal
pay )
.
Processescanbev erysimpl
eorqui
tecomplex.

Inat ypi
calcompany ,
processesmi ghtincludecal cul
ati
ngsal est
rends,
fil
i
ngonlinei
nsuranceclai
ms, or deri
ng
i
nv entor
yfrom asuppl i
er
ssy stem,orv erify
inge-mailaddressesforWebcust omers.Processescont ainthe
businesslogic,al
socal l
edbusi nessrul es,thatt r
ansformt hedataandpr oducet herequi
redr esults.The
symbolf orapr ocessi sar ectanglewi thr oundedcor ner
s.Thenameoft heprocessappearsi nsidet he
rectangl
e.Thepr ocessnamei dentif
iesaspeci fi
cf uncti
onandconsi stsofav erb(andanadj ective,if
necessary)fol
lowedbyasi ngularnoun.Exampl esofprocessnamesar eAPPLYRENTPAYMENT, CALCULATE
COMMI SSI
ON, ASSIGNFINALGRADE, VERI FYORDER, andFI LLORDER.

Forexampl
e,y
oumi
ghthav
eapr
ocessnamedDEPOSI
TPAYMENT.Thepr
ocesssy
mbol
doesnotr
eveal
the

4
busi
nessl
ogi
cfort
heDEPOSI
TPAYMENTpr
ocess.

I
nDFDs,apr ocesssy mbolcanber ef
erredt oasabl ackbox,becauset heinputs,outputs,andgener al
functi
onsoft hepr ocessareknown, buttheunder l
yingdetai
lsandlogicoftheprocessar ehidden.Byshowing
processesasbl ackboxes,ananal ystcancr eateDFDst hatshow how t hesystem f uncti
ons,butav oid
unnecessarydet ai
l andcl
utter.Whent heanal y
stwi shestoshowaddi t
ional
levelsofdetail
,heorshecanzoom
i
nonapr ocesssy mbolandcr eat
eamor ein-
dept hDFDt hatshowst heprocess
sinternalwor ki
ngs whi ch
mightr ev
ealev enmor eprocesses, dataflows,anddat astores.I
nthismanner ,
theinformationsystem canbe
model edasaser i
esofincreasingl
ydet ai
ledpictures.

Anobservercanseecabl
esthatcar
rydat
aintoandoutoft
her
out
er,
butt
her
out
er
sint
ernal
oper
ati
onsar
e
notr
eveal
edonl yther
esul
tsareappar
ent.

DATAFLOW SYMBOL

Adat aflowi sapat hfordat atomov ef rom onepar toft heinformat ionsy stem toanot her.Adataf l
owi naDFD
representsoneormor edat ai t
ems.Forexampl e,adat aflow coul dconsistofasi ngledat ait
em ( suchasa
studentI Dnumber )ori tcoul dincludeasetofdat a( suchasacl assr osterwithst udentIDnumber s,names,
andr egistrat
iondatesf oraspeci f
iccl ass).Alt
hought heDFDdoesnotshowt hedet ail
edcont entsofadat a
fl
ow,t hati nf
ormationisi ncludedi nt hedat adicti
onar y.Thesy mbolf oradat aflow isal inewi t
hasi ngleor
doublear rowhead.Thedat af l
ownameappear sabov e,below, oral ongsidet heli
ne.Adat aflownameconsi sts
ofasi ngularnounandanadj ectiv
e, ifneeded.Exampl esofdat af l
ownamesar eDEPOSI T,I
NVOI CEPAYMENT,
STUDENTGRADE, ORDER, andCOMMI SSION.Except ionstot hesi ngularnamer ulear edataf l
ownames, such
asGRADI NG PARAMETERS,wher easi ngul
arnamecoul dmi sleady ouint othi
nkingasi nglepar ameteror
singleit
em ofdat aexists.Figure5- 5showscor rectexampl esofdat af l
owandpr ocesssy mbolconnect ions.
Becauseapr ocesschangest hedat a
scont entorf orm,atl eastonedat af l
owmustent erandonedat af l
ow
mustexi teachpr ocesssy mbol ,ast heydoi ntheCREATEI NVOI CEpr ocess.Apr ocesssy mbol

5
canhavemor ethanoneoutgoi
ngdat
aflow,asshownintheGRADESTUDENTWORKpr ocess,ormorethan
oneincomingdatafl
ow,asshowni
ntheCALCULATEGROSSPAYpr ocess.Aprocessalsocanconnecttoany
othersymbol,i
ncl
udinganot
herpr
ocesssymbol,asshownbytheconnecti
onbet weenVERI FYORDERand
ASSEMBLEORDERi nFi
gure5-
5.Adat
aflow,t
her
efore,
musthav
eaprocesssymbol onatleastoneend.

Fi
gure5- 6showst hr
eedataflowandpr ocesscombi nat
ionst
haty oumustav oi
d:
Spontaneousgener at
ion.TheAPPLYI NSURANCEPREMI UM process,fori
nstance,producesout put
,buthas
noinputdataflow.Becauseithasnoi nput,
theprocessiscal
ledaspont aneousgenerationprocess.
Bl
ackhol e.TheCALCULATEGROSSPAYi scall
edabl ackholeprocess,whichisaprocesst hathasinput,
but
pr
oducesnoout put.
Grayhol e.Agr ayholeisapr ocessthathasatl eastoneinputandoneout put,butt heinputobv i
ousl
yis
i
nsuffi
cienttogeneratetheoutputshown.Forexampl e,adateofbirt
hinputisnotsuf f
icientt
opr oduceafinal
gr
adeout putintheCALCULATEGRADEpr ocess.

totalear
ningsanddeduct
ionsattheendoftheyear
.ADFDdoesnotshow thedetail
edcontent
sofadat a
store t he speci
fi
c st
ruct
ure and dat
a el
ements ar
e def
ined i
nthe dat
a dict
ionar
y.The physi
cal

6
character
isti
csofadatast
oreareuni
mportantbecausey
ouareconcernedonlywit
hal ogi
calmodel
.Also,the
l
engthoft i
methatthedat
aisstoredisunimport
ant i
tcanbeamat terofsecondswhileatr
ansacti
oni s
processedoraperiodofmonthswhil
edataisaccumulat
edfory
ear-endprocessi
ng.Whatisimport
antisthat
aprocessneedsaccesstothedataatsomelatert
ime.

InaDFD,t heGaneandSar sonsy

mbolf oradat ast
oreisaf l
atr
ectangl
ethatisopenont her i
ghtsi
deand
closedont helef
tside.Thenameoft hedatastoreappearsbet
weent helinesandidentif
iesthedatait
contai
ns.Adatastorenamei sapluralnameconsist
ingofanounandadject
ives,i
fneeded.Examplesofdat
a
storenamesareSTUDENTS,ACCOUNTSRECEI VABLE,PRODUCTS,DAILYPAYMENTS,PURCHASEORDERS,
OUTSTANDI NG CHECKS,I NSURANCEPOLI CI
ES,andEMPLOYEES.Except i
onst othepluralnamer ul
eare
coll
ecti
venounsthatrepr
esentmulti
pleoccurr
encesofobj
ects.

Forexampl e,GRADEBOOKr epresentsagr oupofst udentsandt hei

rscores.Adatast or
emustbeconnect edt o
aprocesswi t
hadat afl
ow.Figure5- 7il
lustr
atestypicalexamplesofdat astor
es.Ineachcase,t hedatastore
hasatl eastonei ncomingandoneout goingdataf l
owandi sconnectedtoapr ocesssymbolwi thadataflow.
Viol
ationsoft herulethatadatast oremusthav eatleastonei ncomingandoneout goi
ngdataf lowareshown
i
nFi gure5-8.Int hefir
stexampl e,twodat astoresar econnectedincorr
ectl
ybecausenopr ocessisbetween
them.Al so,COURSEShasnoi ncomi ngdat afl
owandSTUDENTShasnoout goi
ngdat aflow.I
nt hesecondand
thi
rdexampl es,thedatastoreslackei t
heranout goi
ngori ncomingdat afl
ow.

Thereisanexcept
iontotherequi
rementt
hatadatastor
emusthaveatleastoneincomingandoneoutgoing
datafl
ow.Insomesituat
ions,
adatastor
ehasnoinputdataf
lowbecausei
tcontai
nsf i
xedref
erencedat
athat
7
i
snotupdat ed byt hesystem.Forexample,considera data st
orecall
ed TAX TABLE,which contai
ns
withholdi
ngt axdatathatacompanydownl
oadsfrom theInt
ernal
RevenueServi
ce.Whenthecompanyr unsi
ts
pay r
oll
,theCALCULATEWI THHOLDINGprocessaccessesdatafrom thi
sdatastor
e.OnaDFD,thiswoul dbe
represented as a one-wayoutgoi
ng dat
af low fr
om t he TAX TABLE data st
oreint
ot he CALCULATE
WI THHOLDI NGpr ocess.

ENTITYSYMBOL
Thesy mbolf oranent i
tyisar ectangle,whichmaybeshadedt omakei tlookt hree-dimensional .Thenameof
theent i
tyappearsi nsidet hesy mbol .ADFDshowsonl yext ernalentiti
est hatpr ovidedat at ot hesy stem or
recei
veout putfrom t hesy st em.ADFDshowst heboundar i
esoft hesy stem andhow t hesy stem i nterf
aces
withtheout si
dewor ld.Forexampl e, acustomerent i
tysubmi tsanor dert oanor derprocessi ngsy stem.Ot her
exampl esofent it
iesi ncludeapat i
entwhosuppl iesdat atoamedi calr ecor dssy st
em,ahomeownerwho
recei
vesabi l
lfrom aci t
ypr opertyt axsystem,oranaccount spay ablesy st
em t hatreceivesdat af rom t he
company spurchasi ngsy stem.DFDent i
ti
esal soarecalledtermi nators,becauset heyar edat aor i
ginsorf i
nal
desti
nat i
ons.Systemsanal ystscal lanent it
yt hatsuppliesdat at othesy stem asour ce,andanent it
yt hat
recei
ves dat afrom t he sy stem a si nk.An ent i
tyname i st he singularf orm ofa depar tment ,out side
organizati
on,otheri nformat ionsy stem,orper son.Anext er
nalent it
ycanbeasour ceorasi nkorbot h,but
eachent it
ymustbeconnect edtoapr ocessbyadat af l
ow.Fi gures5- 9and5- 10showcor rectandi ncorrect
exampl esoft hi
sr ule.Wi thanunder st
andingoft hepr operuseofDFDsy mbol s,youar er eadyt oconst ruct
diagr
amst hatuset hesesy mbol s.

Ex
ampl
eofDFD
Ex
ampl
efi
gureshowshowt
hesy
mbol
scanbecomposedt
ofor
m aDFD.TheDFDdescr
ibest
hear
it
hmet
ic
8
expr
essi
on.
(a+b)*(
c+a*d)

Assumi
ngthatthe.dat
aa,b,canddarereadfr
om ater
minalandt
heresul
tinpri
nted.Thef
igur
eshowst
hat
ar
rowcanbeforkedt
orepr
esentt
hef
actthatt
hesamedatum i
susedi
ndiff
erentpl
aces.

Exampl e.Fi
g.descr i
besasimpli
fiedinformationsy stem f
orapubl i
cli
brary.Thedat
aandf uncti
onsshownar e
notnecessar i
l
ycomput erdataandcomput erfunct i
on.TheDFDdescr i
besphy si
calobj
ects,suchasbooksand
shelves,togetherwithdatastoredthatar el i
kel ytobe,butar enotnecessaril
y,r
eali
zedascomput erf
il
es.
Getti
ngabookf rom theshelfcanbedoneei therautomatical
l
y-byar obot-
ormanually.Inbothcases,the
acti
onofget ti
ngabooki srepresentedbyaf unct i
ondepictedbyabubbl e.Thefi
gurecoul
dev enrepr
esentthe
organizati
onofl i
brar
ywithnocomput er
izedpr ocedures.

Remembert hataDFDi snotaf l

owchart.Allnamesshoul dbeuni que.Processesarealwaysrunni
ng, t
heydo
notstartorst
op.Alldataf l
owsarenamed.Keepanot eofal lthepr ocessesandext er
nalent
it
ies.Giveuni
que
namest othem.identi
fythemanneri nwhicht heyi
nteractwitheachot her
.Donumber ingofprocesses.Avoi
d
complexDFDs( i
fpossi bl
e)TheDFDshoul dbei nt
ernall
yconsi stent.Ever
ypr ocessshoul
dhav emi ni
mum of
oneinputandoneout put.Onlydat
aneededt operformt hepr ocessshoul dbeani nputtotheprocess.The
di
recti
onoffl
owi sfr
om sour cetodest
inati
on.

DATADI
CTI
ONARY

9
AsetofDFDspr oducesal ogicalmodeloft hesystem,butt
hedetailswi t
hinthoseDFDsar edocument ed
separatelyinadat adi ct
ionary,whi chist hesecondcomponentofst ructuredanalysi
s.Adatadi ct
ionar
y,or
datar epository
,isacent ralst
or ehouseofi nfor
mationaboutthesystem sdata.Ananalystusest hedata
dict
ionaryt ocollect,document ,andor ganizespecifi
cfact
saboutthesy stem,incl
udingthecontent
sofdat a
fl
ows,dat ast ores,entiti
es,andpr ocesses.Thedat adicti
onar
yalsodefinesanddescr i
besalldataelements
andmeani ngfulcombi nationsofdat aelement s.

Adatael ement,alsocall
edadat ai t
em orfiel
d,i
st hesmal lestpieceofdat athathasmeaningwi thi
nan
i
nfor
mationsy stem.Exampl esofdat aelementsarestudentgr ade,salary,Soci
alSecuri
tynumber,account
bal
ance,andcompanyname.Dat aelementsarecombinedintor ecords,alsocall
eddatastr
uctur
es.Arecordis
ameaningfulcombi nat
ionofrelateddataelementsthati
si ncludedi nadat afloworretai
nedinadat astor
e.
Forexample,anaut opartsstorei nvent
oryrecor
dmi ghti ncludepar tnumber ,descri
pti
on,suppl
iercode,
mini
mum andmaxi mum stocklevels,cost
,andli
stpr
ice.

Si
gnif
icantrel
ati
onshi
psexi stamongtheitemsinadat adict
ionar
y.Forexampl
e,datastor
esanddataf l
ows
ar
ebasedondat astr
uctures,whi
chintur
nar ecomposedofdat ael
ements.Dat
aflowsareconnect
edtodata
st
ores,enti
ti
es,andprocesses.Accur
atel
ydocumenti
ngt heserel
ati
onshi
psisessent
ialsot
hedatadict
ionar
y
i
sconsistentwi
ththeDFDs.

Dat
aDi
cti
onar
yRepor
ts

Thedat adict
ionaryser vesasacent ralstorehouseofdocument at
ionf orani nformati
onsy stem.A dat a
dicti
onaryiscreatedwhent hesy stem i sdeveloped,andi supdat edconst antl
yast hesy st
em isimplemented,
operated,andmai ntai
ned.Inaddi tiont odescribi
ngeachdat aelement ,dataflow,datastore,r
ecord,ent
ity
,and
process,thedat adict
ionarydocument st herel
ationshipsamongt hesecomponent s.
Youcanobt ai
nmanyv aluablerepor tsfrom adat adicti
onary,i
ncludingt hefol
lowing:
Anal phabeti
zedl i
stofalldatael ement sbyname
Ar eportdescribi
ngeachdat ael ementandi ndicati
ngt heuserordepar tmentthatisresponsi
blefordataentry
,
updat i
ng,ordeleti
on
Ar eportofalldatafl
owsanddat ast oresthatuseapar t
icul
ardat aelement
Det ail
edreportsshowi ngal lchar acteri
sti
csofdat aelements,r ecords,dataflows,processes,oranyot her
selectedit
em storedint hedatadi ctionary

DECI
SIONTABLE

Somet imes,itisconv eni

enttodescri
beasy st
em asasetofpossi blecondi ti
onssatisfiedbyt hesystem ata
giventime, r
ulesforreacti
ngtostimuliwhencer t
ainsetsofthosecondi ti
onsar emet ,andact i
onstobet aken
asar esult.
Decisiontablesprovideamechani sm forrecordi
ngcomplexdecisi
onl ogic.Decisiontabl esarewidelyusedin
datapr ocessingapplicat
ionsandhav eanext ensiv
elydevel
opedliterature.Asi l
l
ustratedi nTableadeci si
on
tablei
ssegment edintofourquadr
ants: condit
ionstub,
condi
tionentry,actionstub,andact ionentr
y.

Adecisi
ont ableisalogicalst
ruct
urethatshowsev er
ycombinati
onofconditi
onsandoutcomes.Analyst
s
oft
enusedeci si
ontablestodescri
beapr ocessandensur
ethattheyhaveconsi
der
edallpossi
blesi
tuat
ions.
Youcancreatedecisi
ontablesusi
ngMicrosoftPower
Point
,Word,orExcel
.

TABLESWI THONECONDI TIONIfapr

ocesshasasingl
econdi
ti
on,thereonl
yaretwopossi
bili
ti
esyesorno.
Eit
herthecondi t
ionispresentoriti
snot,sother
eareonlytwor ul
es.Forexample,totr
iggeranover
ti
me
calcul
ation,theprocesscondi
ti
onmightbe:Arethehoursgreat
ert han40?Ifso,thecal
culati
oni
smade.
Otherwise,iti
snot.

TABLESWITH TWO CONDI TIONSSupposey ouwanttocr eat

eadeci si
ontablebasedont heVer i
fyOrder
busi
nesspr
ocessshowni nFigur
e5-34.Whendocument i
ngapr ocess,iti
simpor t
anttoensurethatyouli
st
ever
ypossi
bil
i
ty.I
nt hi
sexample,theprocessdescri
pti
oncontainstwocondi t
ions:pr
oductstockstat
usand
cust
omercr
editst
atus.I
fbothcondit
ionsaremet,t
heorderisaccepted.Other
wisetheorderi
sreject
ed.

10
Thecondit
ionstubcontainsallofthecondi t
ionsbeingexamined.Condi
ti
onentr
iesareusedtocombine
condi
ti
onsint
odecisi
onrules.Theacti
onst ubdescr
ibestheacti
onstobetakeni
nresponset
odeci
sionr
ules,
andtheact
ionentr
yquadrantrel
atesdeci
sionrulest
oactions.

Tablei l
lustrat
est hef ormatoflimi ted-entrydecisiont abl e(entr
iesar el
imitedt oY,N,
,andX) .Inal i
mi t
ed-entrydecisiont abl
e, Ydenot ed y es,Ndenot es nodenotes dontcare,
andXdenot ed
perform act i
on.Accor di
ngt oTabl e,ordersareappr ov edifthecreditlimitisnotexceeded, orift
hecreditli
mit
i
sexceededbutpastexper i
encei sgood,ori faspeci alarrangementhasbeenmade.I fnoneoft hese
condi ti
onshol d,theor derisrej
ect ed.
The( Y,N,)ent ri
esi neachcol umnoft hecondi tionent ryquadr antfrom adeci sionr ul
e.Ifmorethanone
decisionr uleshasi dent i
cal(Y,N,)ent ri
es,thet abl ei ssaidtobeambi guous.Ambi guouspairsofdecision
rulest hatspeci fyident i
calactionsar esaidt ober edundant,andt hosespeci fyingdi f
fer
entactionsar e
contradi ct
ory.Cont radictoryrul
esper mitspeci fi
cat ionofnondet er
mi nisti
candconcur r
entacti
ons.Tabl e
i
llustratesredundantr ul
es( R3andR4)andcont radictor yrules(R2andR3, andR2andR4) .

Advant
agesofDecisi
onTable
Thevar
iousadv
antagesofdeci
siont
abl
ear
edepi
ctedasf
oll
ows:
11
(I
)Consistencyindecisionsmaking
Deci
sionrulesareclearlystr
uctur
edDocument ati
oniseasil
yprepar
ed,
changedorupdated
Managerscanber el
ievedfrom decisi
on-
making
Easytouse
Faci
li
tat
emor ecompactdocument ati
on
Communi cationiseasierbetweenmanagerandanal ysi
s
Easi
ertofollowapar ti
cularpathdownonecol umnthant hr
oughcompl
exandlengt
hyflowchar
ts.
Easi
ertodr awormodi fyincompar i
sontofl
owchar t
s.

DisadvantagesofDeci si
onTabl e
Thev ar
iousdisadvantagesofdecisiont
abl
ear
easfoll
ows:
Doesnotdepi cttheflow
Cannotl istal
l t
healternat
ives
Noteasyt otranslat
eitImposeanaddi
ti
onal
burden

Decisi
onTrees
Adecisiontreeisagr aphi calr
epresentati
onoft heconditi
ons,acti
ons,andr ul
esfoundinadecisi
ontable.
Decisi
ontreesshowt hel ogicst
ructureinahor i
zont
alformthatresemblesat r
eewi t
htheroot
satthelef
tand
the br
anches tothe r ight.Like flowchart
s,decisi
on trees are usefulway sto pr
esentthe sy
stem to
management .Deci
siont reesanddeci siontabl
esprovidethesamer esul
ts,butindif
fer
entfor
ms.Inmany
sit
uati
ons,agraphi
cist hemostef fectivemeansofcommuni cat
ion.

Adecisi
ontreeisreadfr
om lef
ttori
ght
,withthecondi
ti
onsalongthevar
iousbranchesandtheact
ionsatthe
f
arri
ght.Becauset heexamplehastwoconditi
onswit
hf ourresul
ti
ngsetsofacti
ons,theexampl
ehasf our
t
erminat
ingbranchesatther
ightsi
deofthetr
ee.

Whethert
ouseadecisiont
ableoradecisiontr
eeof t
enisamatterofpersonalprefer
ence.Adecisi
ont
abl
e
mightbeabet
terwaytohandlecomplexcombinati
onsofcondi
ti
ons.Ontheot herhand,adeci
siontr
eei
san
ef
fecti
vewayt
odescri
bearelat
ivel
ysi
mpleprocess.

12