scieee Science in your language
[en] (orig)

Usage of architectural styles and technologies in IT companies and organizations

Abstract

As in many other software engineering activities, Non-Functional Requirements (NFR) are fundamental in the process of selection of the architectural style and the technologies for developing a software system. To know more about this issue, we are currently driving a survey to find out which architectural styles and technologies are being in use in IT companies and organizations, and their relation to types of NFR. We want to apply this knowledge to Model-Driven Software Development (MDSD), and in fact we are currently developing a framework that considers NFR in the MDSD process.

Read accessible full text

Usage of architectural styles and technologies in IT companies and organizations

Author: Ameller, David,Franch Gutiérrez, Javier
Year: 2009
Source: https://upcommons.upc.edu/bitstream/2117/8851/1/easa09-da-xf-def.pdf
Usage o a chi ec u al s yles and echnologies in IT
companies and o ganiza ions
Da id Amelle and Xa ie F anch
Uni e si a Poli ècnica de Ca alunya (UPC)
{damelle , anch}@lsi.upc.edu
As in many o he so wa e enginee ing ac i i ies, Non-Func ional Requi emen s (NFR) a e
undamen al in he p ocess o selec ion o he a chi ec u al s yle and he echnologies o
de eloping a so wa e sys em. To know mo e abou his issue, we a e cu en ly d i ing a su ey
o ind ou which a chi ec u al s yles and echnologies a e being in use in IT companies and
o ganiza ions, and hei ela ion o ypes o NFR. We wan o apply his knowledge o Model-
D i en So wa e De elopmen (MDSD), and in ac we a e cu en ly de eloping a amewo k
ha conside s NFR in he MDSD p ocess.
The su ey is a ailable in he ollowing URL: h p://www.lsi.upc.edu/~dsdm-su ey.
De elopmen o he su ey: The su ey has been de eloped ollowing an i e a i e
me hodology. A each i e a ion, i has been e ised by IT expe s and esea che s o he a ea.
The e ha e been h ee i e a ions, he i s one in plain ex and he o he wo using he cu en
elec onic o ma . I was di icul o ind good (and cheap) so wa e o de eloping su eys;
inally, we chose LimeSu ey, which is ee open sou ce so wa e (FOSS).
Fo he dissemina ion o he su ey we a e using mainly wo s a egies. On he one hand,
pe sonal con ac wi h so wa e a chi ec s ha we know pe sonally. On he second hand,
ad e isemen in IT communi ies hos ed in common si es such as LinkedIn and Facebook. In
his i s s age, we ha e con ac ed 7 so wa e a chi ec s and ad e ised in he In e na ional
Associa ion o So wa e A chi ec s g oup and we plan o do he same in o he simila g oups.
Design o he su ey: The su ey is di ided in o h ee pa s: i s , common ques ions abou
so wa e de elopmen ; second, ques ions abou he deg ee o desi ed in e ac ion wi h
de elopmen ools; and hi d, ques ions abou he usage and knowledge o MDSD. The hi d
pa o he su ey is d i en in h ee di e en ways depending on he knowledge o he
esponden s.
In he i s pa o he su ey we ha e asked abou he used a chi ec u al s yles, he ype o
de eloped applica ions, he echnological s yles used in hem, ques ions abou he use o da a
base managemen sys ems (DBMS), and inally ques ions abou NFR.
The second pa consis s o wo ques ions abou he desi able in e ac ion o a hypo he ical
de elopmen ool. The i s ques ion is abou code gene a ion and he second is abou design
decisions.
Finally, in he hi d pa he su ey may ake h ee di e en pa hs. I he esponden doesn’
know he concep o MDSD, he su ey inishes. I he/she knows he concep bu doesn’ use
MDSD in his/he p ojec s, we ask abou his/he knowledge abou MDSD, namely abou he
known ini ia i es, de elopmen amewo ks, and se e al ques ions o ind ou his/he opinion
abou MDSD. I he/she is using MDSD in his/he p ojec s, we ask abou he usage o MDSD. I
is in e es ing o see in he hi d case i hey a e using di e en a chi ec u al s yles o
echnologies when hey de elop so wa e using MDSD.
Cu en s a us: We ha e, a he ime o w i ing, 56 esponses o he su ey. We ha e planed o
wai un il we ha e a leas 100 esponses be o e he publica ion o he esul s. As a p e iew, we
ha e seen he ollowing endencies in he cu en esponses.
F om he a chi ec u al poin o iew, mo e han he 65% o esponden s use 3-laye ed o MVC
a chi ec u al s yles ollowed by clien -se e (51%) and SOA (46%). The mos common ype o
applica ion is Web-based (75%) ollowed by Web Se ices (58%).
0
5
10
15
20
25
30
35
None Ma ginal Medium Impo an C i ical No answe
Main ainabili y
Reusabili y
E iciency
Reliabili y
Usabili y
Po abili y
Cos
S anda ds compliance
O ganiza ional
Fig. 1: impo ance o di e en ypes o NFR in so wa e p ojec s
F om he echnological poin o iew Ja a-based echnologies a e used by a 75% o
esponden s. Only 21% decla e o be using .Ne -based echnologies, which looks a bi s ange
because many so wa e de elopmen companies cu en ly announce ha hey de elop in .Ne
(may be his ac will be educed on he inal esul s), and he leas used is s ack solu ion (e.g.
LAMP) wi h 14%. JSP is he mos used Ja a echnology (57%), bu also a signi ican amoun o
esponden s use S u s (48%) o Se le s (48%).
When we alk abou DBMS, ela ional da abases a e, as expec ed, he mos used (88%), being
O acle (69%), SQL-Se e (33%), and MySQL (32%) a e he mos widesp ead b ands. I is a bi
su p ising ha MySQL is used only by 32% o esponden s while 77% a e de eloping Web-
based applica ions. The only DBMS- ela ed mechanism ha has a ele an amoun o
impo ance is s o ed p ocedu es (39% o esponden s hink ha s o ed p ocedu es a e impo an
o e en c i ical in hei applica ions), whils igge s and checks ha e e y li le impo ance
(80% app oxima ely o esponden s epo ha hese mechanisms ha e no o li le impo ance
o hem).
The impo ance o NFR is no e y clea , we ha e ound a di ision o opinions: while 96% o
esponden s conside NFR (73% a he same le el as unc ional equi emen s), only 57% use
NFR o ake a chi ec u al and echnological decisions. In he Fig. 1 we ha e summa ized he
epo ed impo ance o each indi idual NFR. We can see ha he equi emen s such as
main ainabili y, eusabili y, e iciency, eliabili y, and usabili y ha e a endency o being mo e
impo an han po abili y, cos , s anda d compliance, and o ganiza ional NFR.
When we pu NFR in o p ac ice, 80% o esponden s do no use de elopmen ools ha analyze
he NFR compliance, bu 70% would like o use such kind o ools.
In he second pa o he su ey (ques ions abou in e ac ion) he mos accep ed answe o all
ques ions is ha de elope s wan o be asked, bu only o he mos impo an decisions. This
answe can be ela ed wi h he iew o MDSD no as a ully au oma ic p ocess bu as an assis ed
one, in which some decisions a e aken by he expe .
A g ea pa o he esponden s (79%) do no use MDSD in hei so wa e p ojec s, so i is
di icul o ca y ou a good e alua ion o his pa o he su ey a his momen . On he o he
hand, 50% o hem decla e o know he concep . Eclipse EMP seems o be he mos known
pla o m o MDSD and MDA he mos known me hodology.
This was an o e iew o he cu en answe s o he su ey, o he inal epo on i we will do a
deepe s a is ical da a analysis. Fo example, we will analyze o each a chi ec u al s yle o
echnological s yle ( hese e ms a e de ined in he su ey) which a e he mos impo an NFR.
Conclusions: This su ey can be seen as a pa icula ins umen ha add esses some o he
ques ions aised in he EASA’09 c p. Ou posi ion is ha a way o ob ain empi ical e idence
abou he cu en s a e o so wa e a chi ec u es usage in IT companies and o ganiza ions is
asking he in ol ed ac o s.