The good CASE has a result-oriented life-cycle, proper methods and supporting tools
CASE is not just a tool it is about new working principles. It is based on engineering principles and computer support. It includes graphical presentation and user cooperation.
In Cap Gemini we believe Ihat the rcason for CASE is to be found in todays and yesterdays problems in systems development.
Today our desktop computers have a power thaI would have been unimaginable just a few years ago.ln software there is no similar trend.
There is an intense need for a clear improvement in price/performance concerning development of software.
This will putlhe following requirements on sys tcms development: o Increased productivity in systems develop ment o Increased quality in systems development o Increased effectivity in systems development And it requires a new way of working. How do we find the good CASE -the CASE that supports a good way of working - for systems develop ment in the nineties! Considering the rapid development in the CASE area it is today difficult to chose the right way. Our first advice to a company considering CASE tools? Does the problem concern a lack ofI is. What is the problem? Is it about methods or methods for analysis or is it lack of programming productivity? The CASE tools kan improve the effectivity in the working principles. But if there are no defined working principles the first thing to do is to establish that. CASE in itself does not provide defined working principles. o Go for a good working principles. o Remember that the development of CASE has but started. o Look at the total environment. o In Cap Gemini we will go for integrated working principles.
A result-oriented life-cycle, proper methods and supporting tools.
CASE is not just a tool. CASE is much more. CASE is about new working principles. It is based on engineering principles and com puter support. It includes graphi cal presentation and user cooperation. This is the basis for this presentation.
Why is CASE such a current topic? In Cap Gemini we believe that the reason for CASE is to be found in todays and yesterdays problems in systems develop ment. Quite simply our in heritance.
The development of the price/performance ratio in hardware and software. On our own desks we can see the results of an enormous development in hardware. Today our desktop computers have a power that would have been unimaginable just a few years ago. If we com pare the development of power and performance with the decreasing prices the develop ment is even more fantastic.
Today our money buys much more performance than what they did five or ten years ago.
In software there is no similar trend. Of course we are more ef ficient and productive but this trend is not at all as strong. In many cases it is hardly notice guarantee quality in timely able. The 4GLs are not common deliveries, cost and functionality. ly used and in many cases sys How can we do this? We must tems development is done today have competent staff and good as it was ten years ago. There is working principles supported by an intense need for a clear im good tools. provement in price/performance concern ing developmen t of - Systems Maintenance is an ever software. increasing problem for many companies. An ever growing part - The development towards sys of the resources that we have for tems that support the business systems development is occupied operations. We are more and with maintaining older systems. more building systems that will This is all right if the main improve the competitiveness of tenance produces good value fOr our customers. It is no longer its money. But too often main rationalizations and reduced tenance is seen only as a cost and costs but gains in competitive with no payoff. ness that we are looking for. All Cap Gemini has for some time systems are unique. This puts developed a concept that focuses even greater demands on systems on profitability/usability as a development. A delayed relcase steering mechanism for systems could mean that the expected maintenance. This starting point profit for the company never is necessary but not sufficient. shows up - the competitor might We must also create possibilities already be ahead. We as systems for reducing the differences be developers must be able to tween development and mainDICIIIOI OIIIITID ICDIL ( 1"-' I
DECISION ( ( ( tenance. The structure of a sys tem must not deteriorate for every corrective action taken. We should try to reach an, in prin cipal, unlimited technical life cycle for the systems that we build. A good way of working and CASE tools are a prereq uisite.
Requirements All in all this will put the follow ing requirements on systems development: o Increased productivity in sys
tems development o Increased quality in systems
development o Increased effectivity in sys
tems development This requires a new way of work ing. How do we find the good CASE - the CASE that supports a good way of working - for sys tems development in the nineties! That is what we will spend the rest of this presentation talking about.
Working principles - for whom? Who needs CASE? Who needs new working principles? Before we go into detail about what new working principles would look like we should stop to think about these issues.
Cap Gemini will during 89 make a major investment in a new way of working supported by CASE.
The questions put forward earlier in this paper were the ones that we had to start with trying to answer.
Presently Cap Gemini BRA has 400 DP consultants. The new way of working is meant to be used firstly by our own staff but not only. Our 400 staff are work ing together with our customers staff. They are also invited to use our new way of working.
The new way of working is not only meant for the DP-Profes sionals.!! ! With a trend towards systems that support the business the group of people that are really knowledge able about the company business operations is .also a group that will benefit from our new way of working.
Today there is no methods or tools that covers all aspects of systems development. What we can learn from the fact that methods and tools are in focus is that a model must not havc a life of its own. TIle model must be a natural structure for what are the most important issues, methods and tools.
Requirements on a new way of working We were just addressing thaI!
The way of working must be in101111'0 fDIICiPLII... Demands * Integrated * Graphical
presen ta tion * Small portions * PC-based New working principles why? When we are working in our cus tomers projects there are always a demand for new tools and methods. New tools and methods are needed to perform the work well, tools are needed to perform the work quickly and efficiently.
A model for systems develop ment is seldom asked for. Still a model for the systems develop ment process is an absolute nec cessity. It is the backbonc that holds together methods and tools "g,,"d, Tho d M l o p m o ' model, the methods and the tools must be united - even if they are separate and to some extent ex changeable. - A graphic presentation inter face.
We have learned from the first wave of CASE tools - a picture tells more than a thousand words.
If pictures are good in a tool they are of course also good in descriptions of model and methods. We must create a way of working that is based on graphics not just tools.
A form 10 present the way of working in small portions.
Gone me the days of the thick binders. A new way of working must be delivered in small por tions. A new way of working must be packaged for several pur poses, depending of areas of use and demands from different users of it. This means a flexible model for development - we will come back to that laler. - PC-bflsed.
The way of working must be sup portcd by a PC. Almost all of us havc PCs on our desks and we like to work on a Pc. That way we feel more effective. Working principles for a PC - what is that if not a CASE tool? Working principles - with computer support Coming back to my introduction - to Cap Gemini a CASE tool is not everything. Consequcntly it is not enough with just a CASE tool on PC, to say that the work ing principles are PC-based. That is why we have put Cap Geminis new development model on PC with the help of presentation tools or ." as an integrated part of a CASE tool.
The working principles the model- based on results As mentioned earlier a well defined model for systems development is the basis for in tegrating methods and tools.
Cap Geminis model is based on rcsults: We sce systems develop ment as a set of results that are to be produced. A result could in this context be a program description, a functional descrip tion, a data model etc.
The basis for our model is the descri ption of these results.
What are the quality criteria for each result? What is the purpose of the result? Which methods and tools can be used to produce the result?
Tbe DEVELOPIIENT IIODEL - available on PC The res ults ca n be seen from different angles. Cap Geminis result model describes how the results interrelate in result chains and result areas. Let me givc you an example. The resulting dala base depends on the datya model in the result area of Storage.
Cap Geminis decision model describes the results grouped in decision areas.
Earlier development models were often activity oriented. One of the drawbacks of that is that they are less flexible.
Now we look at the results rather than the activities. To reach a cer tain result the activities can varyaccording to project or according to the method or tool used. There for the activities are handled in the project plan and not in the development model. To focus on results also makes quality as surance easier. (reviews etc).
The working principles the methods Some other important aspects of the methods that Cap Gemini uses: o IRM o Easy to learn o Engaging o Aiming for automatisation Tht IIIULT OIIIITID ilODIL
STORAGE USAGE _r===:-~
PROCESSING
QUALITY ~ ( ( C' o
Reduces the difference be tween development and main tenance In conclusion we like to call this way of working Information En gineering. This is not a method in its own right but rather require ments on the methods that we want to use.
The seminar is an important part in our use of methods. It fits well into project work because it provides: concentraton, quick results, a forum for creative
I debates, a natural opening for computer support, fast documentation of basic business operations and a I.. common view for all those involved.
In Cap Geminis way of working we use the semi nar in several circumstan ces, for instance for project start up, data modelling, function modelling, goal setting, project calcula tion, prototyping etc.
Project work is characterized by seminars and breathing spaces. In the same way the number of people engaged in project work varies from a large group during the seminar and smaller task for ces in between. In this way of working a larger group can be better activated. The quality of the work improves and the or ganisation is more devoted to the work.
Is there enough experience built into the method of the CASE tool? Does the CASE tool cover all areas of methodology? Un fortunately the total tool does not yet exist and it is doubtful if it ever will. Perhaps it is not even to be wished for.
We presently lack computer sup port in the following areas (there are however certain differences between suppliers): o Security o Quality Assurance o Routine sketching o Implementation o Showing certain technical
structures o Distributed systems/data
bases o Prototyping (Prototyping in a
Windows system environ ment)
DATAYODa ~~~~~-- ..............I ~ ~.IJ Formal methods apart, but using methods is also to use earlier ex perience. Is it possible to incor porate our own experience into the CASE tool? Of course it is possible. In Cap Gemini we have chosen to incor porate our methods into the tool.
Through the orientation towards results in our model we can describe our methods directly in the tool, with an example throughout the model, with standards etc.
The working principles tools 4GL, CASE, I-CASE How could our business attain a faster development of our productivity in developing software? The solution is often discussed in terms of 4GL, CASE and 1 CASE. Are there any contradic tions in these concepts? Could I they be integrated? 4GL is firstly a tool for im proving programm ing productivity. 4GLs does not mean that the quality of the systems that we develop im proves. When programming productivity is the main pro blem 4G L~ can be an answer.The problem with 4GLs is their lack of stand ardization, limited areas of application and lack of per formance under top load.
CASE adresses the problems of software development from another angle, that is analYSis.
The main issue is to produce th RIGHT system. Not just t produce a system and fast. CASE has helped to open our eyes to a graphical way to presentation.
With CASE it is, for the first time, possible to work stand ardized and methodological in large projects down to the last specification. With CASE redrawing is no major obstacle.
With CASE it is no longer dif ficult to integrate several types of diagrams.
I-CASE (or Integrated CASE) is a further development of CASE.
An integrated CASE tool creates possibilities for combining the advantages of CASE and 4GLs.
Effective analysis and effective programmll1g.
However, I-CASE is just one more step on the way. There is still a long way to go before we have the total tool. Let us think for while what such a tool could be made up of: functions as dcvclopmcnt tools: o Above all a complete diction
ary (reposi tory) o Simulation and prototyping
facilities o Description of logic and pro gram- and data base genera tions (Methods tailored to the com panys needs) o Word processing o Presentation tools Functions for managcmcnt o Integrated with project management tools o Integrated with systems
management tools End uscr functions o Report generators o Query languages What can CASE give us today? We are only in the beginning of a major development phase. But wecan already today note several advantagcs with using CASE tools.
CASE for increased productivity - quality CASE can improve our produc tivity. We do not have to redraw diagrams. We can automatically generate data bases and screens.
The tool can check for consisten cy in away that we could not have done on our own.
Sometimes the improvement in productivity is not to be seen in models
logic llLOCAL • ENCYCLOPEDIA
CENTRAL
ENCYCLOPEDIA
GENERATOR hard figures. The explanation for that is that the quality improves even more. Our customers have for the first time been able to make a complete analysis. This will show. in beller quality now and in the future, during main tenance.
CASE for beller systems main tenance.
Maybe the biggest payoffs will occur in systems maintenance: o By reducing the differencies between development and maintenance. The system structure does not deteriorate when programs are changed since everything is regenerated from the changed specifications. o The number of deliveries during a systems dcvelopment process are red uced. Now we can reach our results succes sively. The end user has greater possibilities of controlling the whole chain of events. This should give us beller systems and beller maintenance. o In analyzing more thorough Iy we wi II find more errors in an earlier stage and there will be less errors to contain in testing and in production. An error detected during main tenance costs many times more to correct than an error detected during analysis.
What should we do today? Considering the rapid development in the CASE ( ( ( , ( area it is today difficult to chose the right way. The fact that the most influential hardware and software suppliers still have nlot shown their strategies and plans does make it easier. How should a company act? Is it better to wait for something even better? Or should one try to be in the front line? Is there other alternatives? Our first advice to a company considering CASE is. What is the problem? Is it about methods or tools? Does the problem concern a lack of methods for analysis or is it lack of programming produc tivity? The CASE tools kan improve the effectivity in the working prin ciples. But if there are no defined working principles the first thing to do is to establish that. CASE in itself does not provide defined working principles.
Lack of effectivity and quality in the analysis is mostly due to problems with methods. There is a lack of methods, methods exists but are not used or methods exists only in the minds of experienced analysts. A CASE tool could work as a catalyzer. A vehicle for implementation of good methods and to for implementing good working principles. CASE can also fail - if it is not founded on a good way of working.
Lack of programming produc tivity does not neccessarily indi cate a need for a complete 1 CASE tool. Cap Gemini is right now performing tests where we compare qual ity and productivity when using a PC work station as opposed to conventional programming.
It is a qucstion of putting the con ventional programming environ ment on a PC in order to gain advantages.
If the task comprises the whole chain from analysis to program ming I-CASE seems to be the choice. But that means not con sidering the costs of it. On top of that I-CASE is a rather untried concept. Who wants to pay for the development that still has to be done? Considering the large investment this group of products is the most sensitive to develop ment. Will the products available today still exist tomorrow? In tegrated CASE on PC, what role will it play in the future? Many questions? answers? Bul... - Go for good working principles. Try to find the strengths and the weaknesses of todays. principles of working. Go for the profitable solutions. Making the largest in vestment does not have to mean reaping the largest profit. - Remember that the develop ment of CASE has but started. There will be ever more new products and ever more new rcleases. Waiting is probably not a good strategi but neither is going for one single solution. - Look at the total environment. What dictionary does the com pany have? How does it work? [s the dictionary an asset or a bur den? Do not let CASE be a bas tard in the technical environ ment! - What effects will IBMs proposal for a new CASE dic tionary have?