Copy igh © 2022 by Au ho /s and Licensed by IADITI. This is an open access a icle dis ibu ed unde he C ea i e Commons A ibu ion License which pe mi s un es ic ed
use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
Jou nal o In o ma ion Sys ems Enginee ing and Managemen
2022, 7(1), 11689
e-ISSN: 2468-4376
h ps://www.jisem-jou nal.com/
Resea ch A icle
Guiding legacy sys ems o e olu ion
Pma E: a case s udy o main enance and enginee ing
And é Mon ei o1, Gonçalo Viei a1 *
1 Uni e si y o A ei o, Po ugal
*
Co esponding Au ho :
and emon ei [email protected]
Ci a ion: And é Mon ei o, Gonçalo Viei a (2022). Guiding legacy sys ems o e olu ion. Pma E: a case s udy o main enance and enginee ing.
Jou nal o In o ma ion Sys ems Enginee ing and Managemen , 7
(1), 11689.
h ps://doi.o g/10.55267/iad .07.11689
ARTICLE INFO
ABSTRACT
Recei ed: 11 Oc . 2021
Accep ed: 31 Dec. 2021
E en hough so wa e change is ine i able, accu a e main enance can ex end so wa e li espan in a sub le way
when bo h budge and ime cons ain s ge in he way o so wa e eplacemen . In he Uni e si y o A ei o, he
p ojec Pma E – a quiz web pla o m c ea ed o encou age s uden s o like Ma h –
eme ged in he ea ly 1990’s and
s acked se e al applica ions o e he decades wi hou majo planning, cleaning o upg ade. This esul ed in a
huge-sized amewo k ha was c ucial o be always a ailable and onli
ne and had high ope a ional cos , leading
o an inc easing amoun o echnical deb . A e 3 decades, he p ojec was s udied, e ac o ed and e u bished,
leading o a s able consis en amewo k eady o e olu ion and so wa e spinou s. This wo k shows h
ow o
manage and enginee solu ions o main ain a legacy sys em and e ol e i e en when ied up o hea y cons ain s.
CCS CONCEPTS: So wa e and i s enginee ing • So wa e c ea ion and managemen • So wa e pos -
de elopmen issues • Main aining so wa e • So wa e e olu ion
Keywo ds: So wa e enginee ing, so wa e main enance, so wa e e olu ion
INTRODUCTION
So wa e sys ems ha a e de eloped specially o an
o ganiza ion ha e usually a long li e ime (Somme ille, 2016).
Such sys ems, de eloped many yea s ago, using echnologies
and in e aces ha a e now obsole e, a e s ill in use and emain
i al o he no mal unc ioning o business and ep esen a
subs an ial in es men o o ganiza ions (Newby, 1994).
Au ho s in (Benne , 1995) s a e clea cha ac e is ics o a
legacy sys em: age (mo e han 10 yea s old), he size (hund eds
o housands o lines o code), an old p og amming language, a
long his o y o main enance ha inc eased he en opy and he
main enance cos s, he ele ance o he mission and he in e nal
domain knowledge.
The solu ion seems simple: so wa e enewal o building
om sc a ch. Bu he e is a signi ican business isk in simply
sc apping a legacy sys em and eplacing i wi h a sys em ha
has been de eloped using ecen echnology (Somme ille,
2016). One can come ac oss se e al issues: he sys em may be
using applica ion/se ice ules ha a e no p ope ly
documen ed elsewhe e; he e a e clien s/p ocesses ha a e
elian on he legacy sys em; legacy sys ems a ely ha e
comple e speci ica ions - du ing hei li e ime usually ha e
undocumen ed majo changes.
Howe e , e en hese la ge and du able sys ems mus
change in o de o emain use ul. Bu upg ading is also usually
e y expensi e. The eal challenge consis s o inding cos -
e ec i e and quali y solu ions and e ol ing hem in o de o
mee new equi emen s. Typical solu ions a ailable in
managing legacy sys ems include o dina y main enance,
e e se enginee ing, es uc u ing, eenginee ing, mig a ion,
w apping and also disca ding (de Lucia e al., 2001).
One o hese examples is he p ojec Pma E – P ojec o
Ma emá ica Ensino (Pin o e al., 2007), which eme ged in 1989
as a e eshing way o lea n Ma h and de elop a ondness o
ma hema ics (Anjo e al., 2005). The concep was simple bu
ho ough: p o ide he s uden s an in e ac i e in e ace wi h
ma hema ical quizzes and combining compe i i eness wi h a
gami ied applica ion – le els, li es, coun down ime . While
“playing”, he s uden s would o ge ha he “game” was based
on Ma h, one o he leas popula science e e (Bha dwa, 2018;
Flick and Lede man, 2003). The objec i e was o lea n while
playing, which was a hi . As an academic p ojec bo n in a
uni e si y, he applica ion was always ee o use, bo h o
eache s and s uden s (Pma e, 2019). Sponso ships we e only
used on main e en s o gi e s uden s p izes.
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
2 / 10
The web applica ion would be a ailable o aining du ing
he whole yea , culmina ing on a la ge physical e en a he
Uni e si y o A ei o, whe e schools and s uden s would
compe e o op places. As he expec a ions e ol ed, he p ojec
has g own la ge in he las decade o an a e age daily use o
5000 use s and a compe i ion a e age abo e 7000 con es an s.
Table 1 shows he s a is ics o he CNC@UA, he la ges
compe i ion, o he las 5 yea s.
Table 1. Fi e-yea s a s o Pma E CNC@UA compe i ions (de Lucia e al., 2001)
Yea # o Schools # o Teams # o S uden s # o Games
2016
204
4482
7308
5850
2017
195
5263
8468
7261
2018
190
6411
8912
8474
2019
192
6122
8565
8076
2020
Pandemic yea
2021
120
4183
4183
5116
F om i s c ea ion o nowadays, he sys em has g own in bo h
size and ea u es wi h he exis ing echnologies, esponding o
he public demand. None heless he sys em was buil in a ime
when ha dwa e capaci y was a mo e expensi e han i is oday
and, consequen ly, e ficiency ook p io i y o e a main ainable
and upg adable sys em. This b ough ine i able consequences
in e ms o sys em deg ada ion, also caused by poo a chi ec u e
and documen a ion which ine i ably inc eases main enance
cos s (Somme ille, 2016).
In 2019 he Pma E amewo k began o sob due o educed
pe sonnel, low budge and lack o main enance. The na ionwide
ecogni ion, he p ojec b and and he 3-decade e o made he
coo dina o s assume a ein en ing s a egy: he e was an
e ec i e need o make a majo change and o e alua e o he
p ojec . The e was a pe cep ion o in insic high alue o he
p ojec , bu i should be o mally analyzed. The ma ix p oposed
by (Somme ille, 2016) in Figu e 1 was used o e alua e he
sys em and i s ue alue chain, leading o a esul high
alue/low quali y.
Figu e 1. Legacy sys em decisional ma ix (Alkazemi e al., 2013; de Lucia e al., 2001; Ransom e al., 1998), (Newby, 1994)
Low quali y means expensi e o main ain: hese sys ems
should be eplaced, i sui able o - he-shel sys ems a e
a ailable, o eenginee ed o imp o e hei quali y. Making a
decision abou e ol ing a legacy sys em should be able o
ensu e an o e all imp o emen o he quali y o he legacy
sys em, and he educ ion o i s main enance cos s.
Ne e heless, he high alue o bo h s uden s and Uni e si y o
A ei o, on educa ion and ma ke ing, d i es he con inui y o he
p ojec and he sys em eenginee ing and adap abili y o
achie e quali y equi emen s.
In his pape , we p esen a case s udy wi hin a uni e si y,
showing how o e alua e he s a e o a legacy sys em in an
academic o ganiza ion and how o implemen a eenginee ed
solu ion ha can deli e a quali y p oduc wi h ex endable
a chi ec u e and low main enance. The pape is o ganized in
eigh sec ions: sec ion Rela ed Wo k p esen s simila wo k,
sec ion Pma E F amewo k desc ibes he exis ing sys em’s logic
and a chi ec u e ollowed by sec ion Analysis and E alua ion
ha uses a me hodology o assess he sys em; sec ion
Implemen a ion desc ibes he asks’ de elopmen , ollowed by
sec ion Resul s ha compila e da a and gains. Sec ion Fu he
Wo k enume a ee la e wo k o do, while sec ion Conclusions
w aps he achie emen s and esul s.
RELATED WORK
The IT indus y has always pu sued a way minimize he cos
o main aining legacy sys ems. Se e al assessmen models, like
(Newby, 1994), ha e been p oposed and implemen ed, bu
because o he in insic p i a e na u e, closed sou ce o
comme cial, mos a e no accessible o iewable. In he end,
academic wo ks a e mo e isible and usually p o ide some
de ails on published wo k.
Rega ding so wa e e alua ion and main enance, he e a e
some case s udies a ailable, like he Umm Al-Qu a Uni e si y
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
3 / 10
ha esea ched o a solu ion o cope wi h he changing
en i onmen wi hou in e up ing he ou ine wo king
ac i i ies (Alkazemi e al., 2013). The case s udy uses a weigh ed
decision-making g id o elabo a e u u e asks bu does no
implemen any change; la e publica ions om same au ho s
only e e o amewo k p ocess and s a egies (Alkazemi,
2014). Ano he is he case s udy ha uses au oma ed analysis o
iden i y dependencies and e ac o code (Ba o a e al., 2014).
The s udy is in e es ing because o he la ge applica ion ange
bu keeps ocus on p og amming echniques. Au ho s in
(G iswold and No kin, 1993) p opose and implemen a model
and a ool o es uc u e code in an e o - ee ans o ma ion; i
concludes ha i should need e idence o cos -e ec i e
ad an ages om eal-case scena ios. On (Ahmad e al., 2021) we
can ead abou esea ch o mode nize legacy so wa e, wi h a
e y special ocus on mobile applica ion; au ho s p esen a
esea ch me hod, a ans o ma ion p ocess and an applica ion
e alua ion.
As o educa ional games, a e y ecen s udy (The and
Usagawa, 2018) analyzes he e ec i eness o online quizzes,
compa ing adi ional es s wi h i ual es s made on Moodle
pla o m(Moodle P y L d, 2021); i concludes ele a ed
a en i eness and highe achie emen s o online es s bu does
no ocus on quiz gami ying. Ano he in e es ing wo k ocus on
online lip lea ning wi h gami ica ion quiz (Zainuddin e al.,
2021); he s udy p esen s e idence o high engagemen o he
pandemic e a, based on su eys and in e iews, bu does no
show how i was implemen ed o a chi ec u e.
THE Pma E FRAMEWORK
The Pma E amewo k was ini ially a simple e minal
applica ion, w i en in C. Back in he ea ly 90’s ha was a
no el y ha wo ked well wi h he ini ial Ma h quizzes. I had a
simple UI wi h ques ions, each one wi h ou ue/ alse op ions
and a sense o le el p og ess. A he ime i was an excellen s a
o he p ojec ’s objec i e, i mo ed he s uden s as hey enjoyed
“playing” wi h Ma h.
Figu e 1. 1990 e minal applica ion
The nex s ep was o make i a ailable ac oss he campus
ne wo k, online. So, a new applica ion was de eloped in ASP
and VB Sc ip , suppo ing In e ne Explo e . In he ea ly 90’s i
was a b eak h ough and local campus compe i ions became
amous and desi able: 2000 s uden s om he whole coun y
ga he ed annually o ind he bes - s uden , eam o school - in
Ma h games.
A ew yea s la e i was pa ched wi h a ew add-ons,
especially wi h Adobe Flash o suppo Ma h symbols and
o mulas. As he yea s wen by, i had g ea accep ance and
s a ed o g ow o mee demands: mo e games, mo e empla es
o games, mo e school yea s in ol ed and mo e cou ses a he
han Ma h i sel . In 2020 i held 14 main quiz ca ego ies om he
3 d o he 12 h g ade: om Ma hema ics o Biology, passing
h ough Po uguese and English bu algo Physics and
Chemis y and so on. In he las yea s i also was ex ended o
assis highe educa ion classes, suppo ing Calculus
assessmen s o ins ance. A module o build diagnos ic es s was
also de eloped, allowing eache s o e ie e impo an
in o ma ion a he s a o he academic yea o le el-up s uden s
wi h hei knowledge.
The egis a ion, pa icipa ion in compe i ions and access o
esul s ha e been cos - ee om he p ojec ’s beginning. A e
egis a ion, each s uden was en olled in a class by he
co esponding eache and had access o all ca ego ies on
his/he yea bu also all ca ego ies below hei age. The aining
quizzes a e a ailable all-yea long and a e pa ame ized models
illed wi h andom alues a he ime o quiz eques , gene a ing
uncoun able di e en ques ions/answe s. The in e ac i e
in e ace encloses a 4-ques ion quiz game-like applica ion –
le els, li es, coun down ime and ou pai s o T ue/False adio
bu ons, as shown in Figu e 3.
Figu e 3. 2019 quiz sc een sample
A e inishing he le els, o he ime is up, he quiz ends and
he use ge s back o his/he pe sonal a ea. The e he/she can
access o p e ious esul s and pe sonal s a is ics. I seems qui e
simple, bu inside he hood he complexi y g ows
disp opo iona ely.
A chi ec u e
Dis ega ding he ea ly yea s om he e minal C
applica ion, he ac ual Pma E pla o m s a ed as a small ASP
web applica ion wi h a small webse e . Simple clien -se e
calls, low usage and an adequa e da abase whe e he
ques ions we e sa ed, we e enough o he ime-being.
As he yea s wen by, use s s a ed o ask o mo e
academic ange and mo e and di e si ied con en - i g ew in
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
4 / 10
size and ea u es, d i en by use s’ eques s. The manpowe
was always cons i u ed by s uden s o esh-g adua es’
s uden s wi h schola ships, which did no con ibu e o a good
planning o s uc u ing. The o he p oblem was ha a e each
de elopmen cycle, usually om 1-2 yea s, de elope s mo ed
along o o he challenges and he nex s a would s a wi h
no solid g ound. Each demand ou pu ed a so wa e isle ha
was buil wi h he echnology ending a ha ime. No
upda es we e made o he exis ing so wa e, leading o a kind
o so wa e pa chwo k. E e y p ojec , excep o admin back-
o ice, lies on a e sion con ol se e Mic oso TFS, which
allows o make e sion con ol, wi h some small upda es o
changes.
A e 3 decades, i led o a complex amewo k, cons i u ed
o 6 applica ions and 1 da abase – wi h se e al ins ances – as
shown on Figu e 4.
Figu e 4. Pma E amewo k a chi ec u e
The end-use has access o he la es news and his pe sonal
a ea, being able o see esul s, s a is ics, aining and
compe ing wi hou changing URL add ess, hand-o e
be ween PHP and .NET applica ions is ul illed wi h simple
HTTP edi ec s.
A he back-o ice, he adminis a o s can manage all kinds
o use da a and quiz esul s bu mus change he URL o
swi ch applica ion and egis e schools and new s uden s;
hey also mus swi ch o ano he pla o m, ModelMake , o
de elop pa ame e ized quizzes and manage
a eas/ca ego ies/ a ge yea s.
All he da a s ays pe sis en o e a SQL da abase (2013),
holding dozens o ins ances om decades; he la es backup
had o e 10GB and some que ies would ake mo e han 2
minu es. The e is also a Py hon daemon ha p oduces
andom alues o ill pa ame e s: on he on -o ice when a
quiz is gene a ed, a andom model is chosen om he
s uden ’s ca ego y and each pa ame e a iable (ques ion
and/o answe s) is sen o he Py hon daemon o be ende ed
wi h eal alues wi hin he p o ided ange, as shown on
Figu e 5.
Figu e 2. Pa ame e ized quiz empla e and 2 empla e conc e iza ions
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
5 / 10
The amewo k wo ks easonably in daily use, i we look
a he upside – se e al echnologies, mos o hem obsole e,
exchanging da a h ough 7 so wa e blocks. The p oblem is
ha he lack o a cohe en , uni ed s a egy o ollow was
decisi e o inc ease main enance o e he yea s. So many
sys ems, wi h many loose-coupled connec ions, become
comple ely unsus ainable. As a simple example, c ea ing a
new compe i ion o 2020 could ake a whole day – i equi es
a p og amme o c ea e ou new SQL DB ins ances wi h new
compe i ions’ IDs and associa e IDs o co esponden models
manually. Ano he example is he e ie al esul s a e he
annual compe i ions: SQL que ies unning on s o ed
p ocedu es can ake o e 8 minu es o each o he 14 sub-
ca ego ies. All esul s a e hen ou pu ed as HTML o he
esul s public page on he ins i u ional websi e. Too much
wo k, wi h no spa e human esou ces o budge and no
chance o pu he amewo k o line - abou 2000 use s a e
online daily and an a e age o app oxima ely 7000 on
compe i ion days.
Acco ding o (Be ande and And ews, 2005), a possible
app oach is o co ec so wa e aul s, make i adap i e o new
en i onmen s, enhance and imp o e e iciency and o make i
p e en a i e, by making s uc u al changes and subsequen
main enance easie . Wi h his in mind, he objec i es o his
wo k we e hen de ined in 4 main goals:
A. E ol e he amewo k o be main enance ee
B. Speed-up esul s’ da a e ie al
C. Enhance use in e ace and design
D. Cleanup unnecessa y so wa e, addons and da abases
The nex s ep was o go h ough each goal, elici a e he
p oblems and ind a solu ion o i .
ANALYSIS AND EVALUATION
We u ilized he amewo k desc ibed ea lie o assess he
majo asks in o de o build a p io i y able o he cu en s a e
o he sys em. The di e en con ex s, oge he wi h hei
co esponding ac o s and s a uses we e aken in accoun . The
objec i e was o iden i y, ca ego ize and p opose
imp o emen s. The in e p e a ion o he alues ob ained ou
o he Pma E a e gi en in a column called “Ca ego y” ha can
de ines ou le els o impo ance om A o D. These le els
desc ibe he impac o he ob ained alues on business
con inui y and sys em adap abili y om he business
pe spec i e equi emen s. Acco ding o he ou goals de ined
in chap e 3, hey we e also spli o i he ca ego iza ion. The
esul s o he analysis can be seen on Table 2.
Table 2. F amewo k p oblems
# P oblem Solu ion Ca ego y
RP1 Sca e ed so wa e ac oss se e al se e s Me ge so wa e on a single se e wi h edundance and
backup eini ializa ion
A
RP2 Obsole e so wa e obs uc ing se e
echnology
De elop modula and ex ensible new so wa e wi h s a e-o -
he-a p og amming languages
A
RP3 New academic yea s equi e ha d code
ope a ions
Re ac o code and build au oma ic s o ed p ocedu es o
upda e au oma ically
A
RP4 Re ie al o esul s ake oo long Reenginee DB, build indexes and op imize SQL que ies B
RP5 HTML ou pu esul s equi e manual ile
inse ion and edi ing
Re ac o code o ansla e SQL ables in o HTML and inse a
gene a e bu on o he use
B
RP6 Obsole e Quiz UI: no use - iendly, haza d
colo s and i only o 640x480 esolu ion
De elop new iendly UI, wi h esponsi e beha io o i all
de ices, including mobile
C
RP7 Obsole e on and back-o ice De elop a new consis en and cohe en design ans e sal o
all so wa e
C
RP8 Obsole e plugins, addons and webse ices:
Flash and VB Sc ip
Iden i y all necessa y and emo e emaining; eplace by s a e-
o - he-a Ja asc ip lib a ies
D
A e he p oblems we e iden i ied, and as manpowe was
sca ce, hey we e p io i ized o allow inc emen al
de elopmen sp in s and show p og essi e enhancemen on
he pla o m. To achie e his, we used he combined
p io i iza ion echnique e e ed by (Be ande and And ews,
2005), whe e equi emen s a e mixed among s akeholde s – in
ou case P oduc and P ojec Manage s. The esul o ha
p io i iza ion is desc ibed below on Table 3.
Table 3. P io i iza ion Resul s. P io i y, P(RPX) = RPCX × WP oduc M + RPCX × WP ojec dM whe e RP is he equi emen
p io i y, and W is he weigh o he s akeholde
# P oblem P oduc
Manage (0.60)
P ojec
Manage (0.40)
P io i y
RP1 Sca e ed so wa e ac oss se e al se e s 0.10 0.15 0.12
RP2 Obsole e so wa e obs uc ing se e echnology 0.10 0.15 0.12
RP3 New academic yea s equi e ha d code ope a ions 0.25 0.15 0.21
RP4 Re ie al o esul s ake oo long 0.10 0.15 0.12
RP5 HTML ou pu esul s equi e manual ile inse ion and edi ing 0.10 0.10 0.10
RP6 Obsole e Quiz UI: no use - iendly, haza d colo s and i only o 640x480
esolu ion 0.25 0.10 0.19
RP7
Obsole e on and back
-
o ice
0.00
0.10
0.04
RP8 Obsole e plugins, addons and webse ices: Flash and VB Sc ip 0.10 0.10 0.10
To al 1 1 1

Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
6 / 10
A e p io i iza ion, he nex phase was o implemen
changes acco ding o iden i ied signi icance. The ocus would
be on p oblems RP6 and RP3.
IMPLEMENTATION
Following he equi emen s’ p io i y, one by one we
s a ed sol ing he p oblems. The i s s ep was he p oblem’s
diagnosis, hen de elop solu ions o sol e hem and
implemen changes. On his e y i s phase RP1, RP2, RP7
and RP8 we e le on hold.
RP3 – New academic yea c ea ion
Expec a ions we e ha he e could be a page on he back-
o ice o c ea e a new academic yea . This was nonexis en ,
hough. The s anda d p ocedu e was o:
A. C ea e new ids o he 14 compe i ions and sub
compe i ions in he DB;
B. Clone he 5 da abase ins ances (shown on Figu e 6);
C. Go o he DB ins ance o he cu en yea and upda e
compe i ion IDs;
D. Open web o m code ile, commen 400 lines o code
and copy hem o new lines and eplace IDs;
E. Upda e he on -o ice applica ion o c ea e new
quizzes using new IDs and new academic yea .
Figu e 3. SQL Se e DB ins ances
The mos use ul p ocedu e would be a web o m on he
back-o ice wi h lis ings (compe i ion, yea s, e c.) and a bu on
o allow he c ea ion and upda e o all a iables. Based on ha
idea, we c ea ed ha web o m and linked all ac ions o i , so i
would igge e e y hing wi hou edi ing a single line o code.
Be o e ha , we changed he DB in o de o a oid ins ance
c ea ions e e y yea ; we needed o adap and inse new
columns in o ables and apply i on he C# code-behind. The
web o m did he ollowing asks:
A. Au oma e da abase ins an ia ion wi h p ope s o ed
p ocedu es and igge s;
B. Gene a e compe i ion IDs and a ach hem o ela ional
ables;
C. Re ac o code o educe cycles and op imize i ;
D. Upda e on -o ice applica ion o allow yea choosing.
These modi ica ions we e made and esul ed in an
ex emely p ac ical - as well as anspa en - app oach o back-
o ice use s, which can be seen on Figu e 7.
Figu e 4. Au oma ed launch o new Academic Yea
RP6 – Obsole e Quiz UI + RP8 – Obsole e plugins
The ASPX web o m was ull o unnecessa y hings: Flash
plugins, old jQue y lib a ies and old ASP.NET con ols. I also
had huge s a e iews behind he page and was only loaded
wi h session da a, e ie ed om he use login and DB.
Figu e 8. New quiz esponsi e UI
I simpli ied design, added esponsi e elemen s ha i
any sc een and a mo e in ui i e and be e use in e ace.
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
7 / 10
RP4 – Re ie al o esul s ake oo long
The pe o mance p oblem ha exis ed on he amewo k
was e y clea – web o ms ha would e u n some kind o
esul could las om 2 o 8 minu es o un, depending on he
que y. Conside ing he scale, 7-10K s uden s, i may no seem
pa icula ly high, bu he conce n eme ged. To sol e he
p oblem, we s a ed o call he que ies di ec ly on he SQL DB,
o disca d ne wo k o connec ion p oblems. The esul s we e
as onishing, he p oblem was indeed wi h he SQL que ies.
We an he SQL Li e Que y S a is ics and he esul s a e
shown on Figu e 9.
Figu e 5. SQL Que y Analyze on Pma E que y
Besides an e oneous cons uc ion o he que ies, wi h
unnecessa y columns o able joins, he DB was no indexed.
The i s s ep was o co ec que ies wi h solely he necessa y
column da a and able joins. This imp o ed esul s and
inc eased pe o mance, elapsed ime emained be ween 10-20
seconds, depending on he que y. We hen c ea ed indexed
iews o he ables om which we expec ed esul s. He e lies
he pu pose, because he esul s can only be e ie ed when
compe i ions a e o e , which means he iew should be one
iew o inal and consolida ed esul s. We c ea ed a igge o
c ea e a iew o he esul s’ able, like he one shown on Figu e
10, and au oma ically c ea e indexa ion.
Figu e 6. SQL igge o c ea e au oma ic indexa ion
This esul ed in que ies wi h elapsed ime be ween 2-12
seconds, which sounded qui e be e han 1-3 minu es.
RP5 – HTML ou pu esul s equi e manual inse ion
Upon he end o compe i ions, he esul s we e ex ac ed
om he DB wi h SQL que ies like he ones e e ed be o e on
5.3. E en hough hey we e e ie ed wi h HTML ags, each
que y ende ed a HTML ou pu which should be copied and
sa ed on a PHP se e , which would ake a manual e o o
hou s. This was co ec ed o a page ha would call he s o ed
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
8 / 10
p ocedu e and sa e HTML inpu o he co ec olde , wi h IIS
pe missions. I some esul s would equi e e ision (i.e. some
e o on some quiz) i would be enough o dele e he ile om
he se e and access he page again o e esh he ou pu .
Again, his esul ed in easonable ime sa ings and a oided
IT pe sonnel o handle he esul s. Figu e 11 shows an
example o he ou pu .
Figu e 11. Gene a ing HTML esul s wi h one-click bu on
RESULTS
The ou come o he ask’s implemen a ion was e alua ed
bo h in e iciency and usabili y, depending on he
equi emen . This sec ion p esen s a compila ion o he e ec s.
Fo ime-consuming asks, we measu ed he cu en imes
agains olde one, al hough 1-day ask o a p og amme was
an es ima e
Table 3. Implemen ed RP e iciency
# P oblem Ini ial
Time
Cu en
Time
Time
Gain Ra io
RP3
Opening new academic yea s 6:00:00 0:01:50 5:58:10 19636%
RP4
Re ie al o esul s o s uden s 0:02:00 0:00:10 0:01:50 1200%
RP4
Re ie al o esul s o schools 0:08:00 0:00:20 0:07:40 2400%
RP5
HTML ou pu esul s equi e manual ile inse ion and edi ing 4:00:00 0:00:40 3:59:20 36000%
Fo he usabili y assessmen , we made a small su ey wi h
5 ques ions on a 20 use se – we aim o make a su ey on he
end o each compe i ion o b oad alida e he esul s wi h a
mode adequa e scale.
# P oblem Ini ial UI Cu en
UI Gain Ra io
RP6
Obsole e Quiz UI: no use - iendly, haza d colo s and i only
o 640x480 esolu ion 44% 92% 48% 209%
The main goal o an algo i hm is o p oduce co ec ou pu ,
and he sys em was p oducing i ; he main achie emen we
aimed was e iciency, which was ex emely op imized wi h
he bes p ac ices on he se e al sys ems.
FUTURE WORK
Due o ime, manpowe and budge cons ain s, some
equi emen s we e s ill pending. In his case, he cen e ing o
he so wa e on a single se e , wi h backup in case o allback
o cou se (RP1), he whole e ac o ing and upda ing o old
Mon ei o and Viei a / J INFORM SYSTEMS ENG, 7 (1), 11689
9 / 10
echnologies (RP2) and he enewal o he en i e back-o ice
(RP7), w i en in ASP and .NET 2.0 did no go h ough.
RP1 was a p io i y bu he big scale o he sys ems’
in ol ed we e di icul o ga he and educe in such a sho
ime. I should be a p io i y and should be conside ed o
de elop wi h new echnologies.
The RP2 ask was also no comple ed due o ime
cons ain s and o IT se ices echnology lockdown. I was no
possible o assemble a new a chi ec u e wi h s a e-o - he a
echnologies while implemen ing he o he equi emen s.
As o RP7, as i was no a p io i y o s akeholde s, i will
be a ask o he nex de elopmen i e a ion. S ill, i needs
some SQL e u bish also because he pages’ con en holds
di ec SQL que ies, as we can see on Figu e 12 below.
Figu e 7. Example o di ec SQL que ies: muda escola.asp ile
The bes p ac ices indica e ha hey should be
ans o med in ansac - eady que ies and isola ed in s o ed
p ocedu es, which will sa e a lo in main enance bu also can
speed up asks.
One ul ima e ask o ul ill should be a use e alua ion
es : ga he a signi ican g oup o use s and es usabili y and
sa is ac ion. This would aise he le el o he applica ion o
he highes s anda ds.
CONCLUSIONS
A legacy applica ion a ely can be op imized wi hou
signi ican e o . I is o en in ol es ha d and complex wo k
ha is no isible. Thus, a legacy sys em has a di e en
li ecycle whe e decision p ocesses mus be ca ied ou
ca e ully in o de o suppo he success ul managemen o
he sys em. These p ocesses a e usually analyzed in an
implici way, bu when he complexi y o he p oblem is no
common, one mus apply a sys ema ic app oach o de ining
which modi ica ions should be made o ecommended.
This pape p esen ed a use case o how o e ol e a legacy
sys em in o de o main ain i s in insic high alue. I s model
was assessed, analyzed and hen i was e u bished. The
business goals o he legacy sys em we e de ined by aking
in o accoun he poin s o iew o di e en s akeholde s
wi hin he o ganiza ion. The amewo k in e en ions we e
ca ied-ou on c i ical so wa e componen s, hus p o iding
a con e sion esul b ough use usabili y, con olled cos s
and easie main enance o he coming yea s. E en hough i
lacks a la ge scale alida ion, wi h end-use es s, i
augmen ed use iendliness, made da a and in e ace
loadings as e o all use s as well as sa ed se e s’
esou ces.
The use case p o ides a e e ence model ha shows a
po en ial solu ion o legacy sys ems managemen , and mo e
speci ically how add ess hem by aking in o accoun bo h
business and echnical issues. Fu he mo e, models,
echniques, and ools o suppo he amewo k ins an ia ion
we e p esen ed and applied.
As his was he e y i s phase o he enewal, we a e
ce ain ha con inuing he bes p ac ices desc ibed in he
cu en li e a u e will esul in a sys ema ic inc ease o
amewo k quali y and alue. In he second phase o he p ojec
we will add ess he emaining p oblems.
Ne e heless, he expe imen was alida ed wi h excellen
esul s. The conclusi e esul s o he de elopmen will be
p esen ed in u u e wo k.
REFERENCES
Ahmad, A., Alkhalil, A., Al amimi, A.B., Sul an, K. and Khan,
W. (2021). Mode nizing Legacy So wa e as Con ex —
Sensi i e and Po able Mobile-Enabled Applica ion. IT
P o essional, 23(1), pp.42-50.
Alkazemi, B., (2014). A F amewo k o Assess Legacy So wa e
Sys ems. J. So w., 9(1), pp.111-115.
Alkazemi, B.Y., Nou , M.K., Meelud, A.Q. (2013). Towa ds a
F amewo k o Assess Legacy Sys ems, in: 2013 IEEE
In e na ional Con e ence on Sys ems, Man, and
Cybe ne ics. pp. 924–928.
Anjo, A., Pin o, J.S., Oli ei a, M.P., Isid o, R.O.G. and Pais,
S.I.V. (2005). Compu e ized Diagnos ic Tes . Cade nos de