m PAJJEHJ s see in ai odi opi RN ; i uzde! osn boat oves kaiaije za ki zbi pi idi a ju ida, k m  % ui ; ši | Ji
— ni on Dao A a a | |

JANUARY 29, 1990

|

"User/Analyst Teams Make the Best Systems Developers

For decades, development. has been done  gree of IS involvement over an orendet within a tightly constrained window of

APPLIED INTELLIGENCE |

This is the by systems analysts using plastic tem- period of time. | time, called a time box. At the end of
sixth in a series ile and paper to create a binder of End users are inočived say in the ini- the time box, the prototype evolves to
on rapid appli- — specifications. The specifications are tial specification phase of the project  , the point that it supports all of the
cations develop- handed over to programmers who do and in the final integration and cutover; functions reguired by the end users.
ment (RAD), a lineby-line coding with a third-genera- phases. During the lengthy process of Cutover includes development of the
development tion language such as COBOL, FOR- design, coding, testing and integration, final production product, final testing, |
methodology de-  'TRAN or PL/I. The code produced re-..' the business environment may change acceptance testing, final training, instal- |
signed to vyield. guires extensive testing and debugging. significantly, invalidating many of the lation of the production-system environ:
much faster re- Methodologies were created because original functional reguirements. ment, data Conversion, moving the SYS-
sults than tradi- of the difficulties inherent in the tradi: The deficiencies of the traditional life-. tem into production and reviewing sys-
tional methods. tional process. However, in much of IS: / cycle process can be resolved by the tem operation. End users are heavily h
RAD is more development, the methodology is the" ./ RAD methodology. The RAD life cycle involved in all aspects of the effort.
successtul than problem. 4. for an application with a complexity of The RAD life cycle has major advan:
the traditional Usually it consists of a set of paper- ' | about 1,000 function points is relatively | tages over the traditional life cycle.
life-cycle process work binders that establish a rigid set '—" short, as shown in the figure. These include the following:

because it deliv- of tasks to be carried out bi developers. ; The JA em peto is short
' ers hist: -guality sdlutions faster. Just as

important, the RAD life cycle is de-

€ Speed: RAD produces results very
guickly, with high-guality design and
implementation.

signed to meet users" needs as effective- MM 6 Oualitv: This means meeting the t
Jo AE bine ino the System i The Promise of Rapid Application Development: ee lede da etIeb VENE DANIMI
implemented.

Fastet Production, Less Manpower, Higher Ouality | the time the system is cut over.
€ Avoidance of programming by
Cornparison of Traditional and RAD Life Cyeles. |. | |. hand: Code is generated automatically
(ideally on the desktop) for an entire
application by the integrated generator
rem jan? of the CASE tool.
go En —— Ee HSV | (€ Better communication with end us-
KE Hi ers: The knowledge of users is harnessed
and reguirements are communicated
through joint a pplications-developmen t

The RAD life cycleis fundamentally
different from the traditional develop-
ment life cycle. It combines many of the
technigues found to be most; successful
in building systems fast. .

These include rapid prototyping, reus-
able designs, end-user workshops for

planning and design, integrated comput-
er-aided sofuware engineering (CASE)
lools, code generators and industry-stan-

|

[S | HE | workshops. A CASE tool is used within
| "dard repositories. VE vi tj the workshops to capture design specifi-
| Other technigues found to be SuCCESS- pe O ET g cations and build the prototype.
| mi Are aulomatic rule-based validation 2 pari a a imi ulradii z Hi; e Enterprisewide planning: RAD incor-
| and c« lordination, data modeling, process Š ti ij ei jin al testing a" TESLA sor NI HI — porates technigues to define an
| modeling, joint applications-design tech- Sci [o glč sme sina otriewy ,— .. NEMIMNIH enterprisewide strategy, planning, data
| nMagues, highly motivated small teams IS: | leh ope sd applica- - | lji il ji uže ii modeling and process modeling. Formal
| a o elive management, processes. 8h mk Ele Ko wi it other . i tos HU interaction with high-level users encour-
hese technigues are individually MSe| Moe JE lo Moe ug M pozi poe zi že (od
| valuable if used correctly. In synergistic H K4' (m ME V | PE sije pri 4 zoži gi GA! ore ta REED DE MENJA
combinations, they become more power- M4 H"MNM' | —liw | o ue
| (ul, resulling in a revolutionary change A KE A SLI
tax and internal errors in semantics are
| in (he development of systems. $ii caught automaticall by the CASE tool
| Modern development methodologies | Abu analyzer. J fore š
| suc ch a5 na combin Kaabe, nts - coin geo |
RES mm gesta SE se
| | are stored in the repository of a E |
l k proc SS. vab m eh k s. E
| Tool sets have evolved to support oh eh en btcea: tije integrity and
| lhese combinations with a maximum ama ei of the specifications.
| amount of automation. The RAD meth: Reusability: Object-oriented planning |
| odology guides (he practilioners in the an design sre Incorporated to Behjeve pova
best. use of he combined technigues. PERE DEJ a the maximum degree of reusable data
Vhe traditional life-cycle process has (ca O KREATIN EI NU E. structures, procedures, components, tem-
serious deficiencies. Users cannot obtain HO jige.£ HERU4 pit si iu | ml i kij trem baesana designa ME mi Ge. o Bem
stralegically important applications | | | 4 BS [, state code

when they need them. There's often a Unlike the classic life cycle, the RAD

a vine lavi Methodologies were created because of to di fles | UVE, very sage ob tele.
uire- NJA OM m zno
enih Opatšmni Gost neh mega to a inherent in traditional development methods. But in much deo The RAD life cycle is designed to
aloe a aajaja TE VRATI, of IS do opne, the methodology is the problem | Ie Too
| professlonals do not work well together ——— aii DET out tre developenenk of ina oratea Co) |... Zlo
dini lea bonal process b User have i [ra FtošoE Sali oi, $ i; sures that their needs will be met. when ALJO dio
by IS on ke te users pč ni le (| unlikely to change substantill during "8 ot the system" 6 MERO ŠE | EI
ie ni prak lo check and are Mepla, nik methodology a as- the RAD life cycle. End users are active- | Next week's column disene« Mm | Gi
Au o£ inconsiste R/

ziiča sume (hat the developers do not under. ' 'ly involved in all phases PR ije pete
es eko an er«.'— stand how to develop an application ment effort. be odravea Ce ni | piasea balhaia

—rors, In addition, regul :
on, reg and must be told, step by step, what to | - system design are performed by small ' The

change after the specific: s: Hi psov 8

frozen. , teams of end users and IS analysts ke
Because of the long delay in btanie eh reality, the best development is of- ' joint applications-development aaa nI | oro ze - moneta ar James.
parola systems seen as important to the know ezacily What to do Ta nara ir pi zrejo the guidance of a work- — " mation on this volume, call (800) 292. ča
nis "menu | s j
oporni esem as a botlenek (nt pre. podila polov 8 slov and rigid |" During ihe construction phase, a hgiv | doma (ide Uri Sales and Cm |
vents management from evol can build a... |y trained team of IS States a |
business in ways that noge die its na oval o omak, 1, Prototype of (he a Pplcation using so kJ brin Santa a MOV še |
Competitiveness, life cycle ( figure, the traditional ' CASE tool. As the prototype, evolves, it ' 90402 zi5 sa n onica, Calif. |
The traditional development lif or the development of an ap- ' is evaluated by a small na,.1 94-8305. In Europe, con- |
uses inefficient, outdated tecinlauec, cycle. plication with a complexity of about ' users team of end teči Savant, 2 New St., Car; orth,
1,000 funetion points reguires a high de-  " The construction z— ls af ka (5 9) ona GA United K ingdom |
| |