Software Design

1
Software Design
Introduction
Material drawn from Godfrey96,Parnas86,Parnas94
2
Software Design

• How to implement the what.
• Requirements Document (RD) is starting point.
• Software design is a highly-creative activity.
• Good designers are worth their weight in gold!
• Highly sought after, head-hunted, well-paid.
• Experience alone is not enough
• creativity, vision, all-around brilliance
  required.

3
Software Design (Contd)

• Some consider software design to be a black
  art
• difficult to prescribe how to do it
• hard to measure a good design objectively
• I know a good design when I see it.

4
Requirements EngineeringAn Overview

• Basic goal To understand the problem as
  perceived by the user.
• Activities of RE are problem oriented.
• Focus on what, not how
• Dont cloud the RD with unnecessary detail
• Dont pre-constrain design.
• After RE is done, do software design
• solution oriented
• how to implement the what

5
Requirements EngineeringAn Overview

• Key to RE is good communication between customer
  and developers.
• Work from Requirements Document as guide.

6
Requirements Engineering

• Basically, its the process of determining and
  establishing the precise expectations of the
  customer about the proposed software system.

7
The Two Kinds of Requirements

• Functional The precise tasks or functions the
  system is to perform.
• e.g., details of a flight reservation system
• Non-functional Usually, a constraint of some
  kind on the system or its construction
• e.g., expected performance and memory
  requirements, process model used, implementation
  language and platform, compatibility with other
  tools, deadlines, ...

8
The Purpose of RE

• Raw user requirements are often
• vague
• contradictory
• impractical or impossible to implement
• overly concrete
• just plain wrong
• The purpose of RE is to get a usable set of
  requirements from which the system may be
  designed and implemented, with minimal
  surprises.

9
The RE Process
Requirements Analysis
leads to
Requirements Definition
produces
Requirements Specification
Software Specification
included in
10
The Requirements Document

• The official statement of what is required of the
  system developers.
• Includes system models, requirements definition,
  and requirements specification.
• Not a design document.
• States functional and non-functional
  requirements.
• Serves as a reference document for maintenance.

11
Requirements Document Requirements

• Should be easy to change as requirements evolve.
• Must be kept up-to-date as system changes.

12
The Requirements Document Should State ...

• Foreseen problems
• wont support Win-3.x apps
• Expected evolution
• will port to MacOS in next version
• Response to unexpected events/usage
• if input data in old format, will auto-convert

13
Requirements Document Structure

• Introduction (describe need for system)
• Functional Requirements
• Non-Functional Requirements
• System Evolution (describe anticipated changes)
• Glossary (technical and/or new jargon)
• Appendices
• Index

14
A Story ...
Dear Mr. Architect, Please design and build me a
house. I am not quite sure of what I need, so
you should use your discretion. My house should
have between two and forty-five bedrooms. Just
make sure the plans are such that bedrooms can be
easily added or deleted. When you bring the
blueprints to me, I will make the final decision
of what I want. Also bring me the cost breakdown
for each configuration so that I can arbitrarily
pick one.
15
A Story (Contd)
Keep in mind that the house I ultimately chose
must cost less than the one I am currently living
in. Make sure, however, that you correct all the
deficiencies that currently exist in my house
(the floor of my kitchen vibrates when I walk
across it, and the walls dont have nearly enough
insulation in them). Also keep in mind as you
design this house that I wish to keep
yearly maintenance cost as low as possible. This
should mean the incorporation of extra-cost
features like aluminum or vinyl siding. If you
chose not to specify aluminum, be prepared to
explain in detail.
16
A Story (Contd)
Please take care that modern design practices and
the latest materials are used in construction of
the house. The house should be really
nice. However, be alerted that the kitchen should
be designed to accommodate among other things, my
1952 Gibson refrigerator. To assure that you are
building the correct house for our family,
make sure that you contact each of the children
and also the in-laws. My mother-in-law will have
very strong feelings about how the house ought to
be designed since she visits with us at least
once a year. Make sure that you weigh all these
options carefully and make the right decision. I,
however, retain the right to override any
decision you come up with.
17
A Story (Contd)
Please dont bother me with small details right
now. Your job is to develop the overall plans
for this house. Get the big picture. It is not
appropriate at this time to be choosing the color
of the carpet. However, keep in mind that my wife
likes green. Also do not worry at this time
about acquiring resources to build this house.
Your first priority is to develop detailed plans
and specifications. However, once I accept these
plans, I will expect to have the house under roof
within 48 hours.
18
A Story (Contd)
While you are designing this house specifically
for me, keep in mind that sooner or later I will
have to sell this house. It should have appeal
to potential buyers. Please make sure that
before you finalize the plans, there is a
consensus of the population in my area that
they like the features this house has. You are
advised to run up and look at my neighbors house
he had constructed last year. We like it a great
deal. It has many features that we would like to
have in our new home, particularly the
75-foot swimming pool. With careful engineering
I believe that you can design this into our new
house without impacting the construction cost.
19
A Story (Contd)
Please prepare a complete set of blueprints. It
is not necessary at this time to do the real
design since these blueprints will be used
only for construction bids. Please be advised
however, that any increase of cost in the future
as a result of design changes will result in
you getting your hands slapped. You must be
thrilled to be working on such an interesting
project such as this. To be able to use new
kinds of construction and to be given such
freedom in your designs is something that doesnt
happen very often. Contact me as rapidly as
possible with your design ideas. I am
enthusiastic about seeing what you can come up
with.
20
A Story (Contd)
P.S. My wife has just told me that she disagrees
with many on the instructions Ive given
you in this letter. As architect it is
your responsibility to resolve these issues. I
have tried in the past and have been
unable to accomplish this. If you cant handle
this, Ill have to look for a new
architect. P.P.S. Perhaps what I need is not a
house at all, but a travel trailer.
Please advise me as early as possible if that is
the case.
21
RE Summary

• RE focuses on determining what the customer
  wants, and not how it will be implemented.
• RE is hard to get correct it requires good
  communication skills.
• Requirements may change over time.
• RE requires iteration.

22
RE Summary (Contd)

• The customer often doesnt have good grasp of
  what he wants.
• Errors made at the requirements stage are very
  expensive to fix later.
• You might well implement the stated requirements
  correctly, but it wont be the system the
  customer really wants.

23
Back to Software Design ...
24
High-Level (abstract) design
Low-Level (detailed) design
25
Top-Down vs Bottom-Up Design

• Top-down Design
• Start with a coarsely-grained view of system, and
  repeatedly refine components until you have
  concrete sub-components.
• Bottom-up Design
• Start with existing components and glue them
  together to get what you want.

26
Top-Down vs Bottom-Up Design (Contd)

• Top-down is the ideal of most design methods,
  but its rarely followed absolutely
• some branches of development are expanded before
  others are even started
• doesnt adequately account for reuse of existing
  components
• COTS products, libraries, previous versions of
  the same system.

27
Design Quality

• Software design quality, as with other ideas on
  quality, is an elusive concept
• It depends on priorities of your company and the
  customers
• fastest to implement
• easiest to implement
• easiest to maintain, evolve, port
• most efficient/reliable/robust end-product.

28
Discussion

• What does quality mean to
• IBM?
• Microsoft?
• Netscape?
• FAA?
• IRS?
• Intel?
• ...

29
Some Desirable Design Attributes

• Hierarchical A good design should be organized
  into a well-designed hierarchy of components.
• Modular Separate distinct concerns (data and
  processing) into distinct containers (i.e.,
  subsystems, modules, and/or classes). Hide
  implementation details and provide clean, simple
  interfaces for each container.

30
Some Desirable Design Attributes (Contd)

• Independent Group similar things together limit
  the amount of special knowledge that unrelated
  components may share. If you change your mind
  about something, the impact will be localized.

31
Some Desirable Design Attributes (Contd)

• Simple Interfaces Endless flexibility adds
  complexity. Complex interfaces mean
• hard to understand by users and developers (e.g.,
  Unix man page syndrome)
• many possible variations of use
• inconvenient to change interface in order to
  eliminate bad options.
• You can get away with flexible interfaces in a
  low-level localized setting, but the larger the
  scale, the simpler the interface should be.

32
A Rational Design ProcessHow and Why to Fake it
Many have sought a software design process that
allows a program to be derived systematically
from a precise statement of requirements.
although we will not succeed in designing a
real product that way, we can produce
documentation that makes it appear that the
software was designed by such a process
D. L. Parnas
33
RDP - Faking It

• The rational design process is an irrational
  ideal.
• Question If we rarely act in a purely top-down
  way when we develop software, why do most design
  methods assume we do?
• Possible answers
• Its simpler than trying to model real-life.
• Wide variation in problems, possible solutions.
• There is utility in the structure of the process,
  and its ongoing documentation.

34
RPD Payoff
The real payoff comes during maintenance and
the next time around.
35
The Role of Documentation

• Documentation plays a major role in the
  development of any large, long-lived software
  system BUT poor documentation is a monumental,
  ubiquitous problem.
• Most programmers regard documentation as a
  necessary evil, written as an afterthought only
  because some bureaucrat requires it. They do not
  expect it to be useful.
• This attitude is a self-fulfilling prophecy!

36
The Role of Documentation (Contd)

• While most documents are incomplete and
  inaccurate, these problems can be fixed easily.
  More serious problems are
• Poor organization.
• Boring, redundant, verbose prose
• Boredom leads to inattentive reading and
  undiscovered errors.
• Confusing and inconsistent terminology.
• Myopia, cant see the forest through the trees.
• Focus is on documenting small details rather than
  on important design decisions.

37
So Whats to be Done?

• Look at the various stages of the Rational
  Design Process, and consider the documents that
  are to be produced at each step.
• Even if you dont follow the steps in that order,
  go back and fill in the blanks! Pretend that you
  did follow the process precisely.
• It isnt just a paper trail youre creating.

38
Programs vs Proofs

• Designing a software system is a lot like proving
  a mathematical theorem.
• The construction of an original proof is a
  painful process you make lots of mistakes,
  pursue bad paths, ...
• However, once youve figured out how to get
  there, you clean up the proof and present it as
  if no mistakes had ever been made.

39
Programs vs Proofs

• If you want to prove a similar but different
  theorem, you can re-use ideas from the first
  proof!
• The main difference is that during software
  development, you also record why you chose each
  path, what alternatives were considered, and why
  other paths were not chosen.

40
Software Aging
Programs, like people, get old. We cant
prevent aging, but we can understand its
causes, take steps to limit its effects,
temporarily reverse some of the damage it has
caused, and prepare for the day when the
software is no longer viable. ... (We must) lose
our preoccupation with the first release and
focus on the long term health of our
products. D.L. Parnas
41
Software Aging?

• It does not make sense to talk about software
  aging!
• Software is a mathematical product, mathematics
  does not decay with time.
• If a theorem was correct 200 years ago, it will
  be correct tomorrow.
• If a program is correct today, it will be correct
  100 years from now.
• If a program is wrong 100 years from now, it must
  have been wrong when it was written.
• All of the above statements are true, but not
  really relevant.

42
Software Does Age

• Software aging is gaining in significance
  because
• of the growing economic importance of software,
• software is the capital of many high-tech firms.

43
Software Does Age

• The authors and owners of new software products
  often look at aging software with disdain.
• If only the software had been designed using
  todays languages and techniques
• Like a young jogger scoffing at an 86 year old
  man (ex-champion swimmer) and saying that he
  should have exercised more in his youth!

44
The Causes of Software Aging

• There are two types of software aging
• Lack of Movement Aging caused by the failure of
  the products owners to modify it to meet
  changing needs.
• Ignorant Surgery Aging caused as a result of
  changes that are made.
• This one-two punch can lead to rapid decline in
  the value of a software product.

45
Lack of Movement

• Unless software is frequently updated, its users
  will become dissatisfied and change to a new
  product.
• Excellent software developed in the 60s would
  work perfectly well today, but nobody would use
  it.
• That software has aged even though nobody has
  touched it.
• Actually, it has aged because nobody bothered to
  touch it.

46
Ignorant Surgery

• One must upgrade software to prevent aging.
• Changing software can cause aging too.
• Changes are made by people who do not understand
  the software.
• Hence, software structure degrades.

47
Ignorant Surgery (Contd)

• After many such changes nobody understands the
  software
• the original designers no longer understand the
  modified software,
• those who made the modification still do not
  understand the software.
• Changes take longer and introduce new bugs.
• Inconsistent and inaccurate documentation makes
  changing the software harder to do.

48
The Cost of Software Failure

• Inability to keep up,
• reduced performance,
• decreasing reliability.

49
Inability To Keep Up

• As software ages, it grows bigger.
• Weight gain is a result of the fact that the
  easiest way to add a feature is to add new code.
• Changes become more difficult as the size of the
  software increases because
• There is more code to change,
• it is more difficult to find the routines that
  must be changed.
• Result Customers switch to a younger product
  to get the new features.

50
Reduced Performance

• As the size of the program grows, it places more
  demands on the computer memory.
• Customers must upgrade their computers to get
  acceptable response.
• Performance decreases because of poor design that
  has resulted from long-term ad hoc maintenance.
• A younger product will run faster and use less
  memory because it was designed to support the new
  features.

51
Decreasing Reliability

• As the software is maintained, errors are
  introduced.
• Many studies have shown that each time an attempt
  is made to decrease the failure rate of a system,
  the failure rate got worse!
• That means that, on average, more than one error
  is introduced for every repaired error.

52
Decreasing Reliability (Contd)

• Often the choice is to either
• abandon the project
• stop fixing bugs
• For a commercial product, Parnas was once told
  that the list of known unrepaired bugs exceeded
  2,000.

53
Reducing the Cost of SW Aging

• We should be looking far beyond the first release
  to the time when the product is old.
• Inexperienced programmers get a rush after the
  first successful compile or demonstration.
• Experienced programmers realize that this is only
  the beginning ...

54
Reducing the Cost of SW Aging (Contd)

• Responsible, professional, organizations realize
  that more work is invested between the time after
  the first successful run and the first release
  than is required to get the first successful run.
• Extensive testing and rigorous reviews are
  necessary.

55
Preventive Medicine

• Design for success
• Keep records (documentation)
• Seek second opinions (reviews)

56
Design for Success

• Design for change.
• This principle is known by various names
• information hiding
• abstraction
• separation of concerns
• data hiding
• object-orientation

57
Design for Change

• To apply this principle one begins by trying to
  characterize the changes that are likely to occur
  over the lifetime of a product.
• Since actual changes cannot be predicted,
  predictions will be about classes of changes
• changes in the UI
• change to a new windowing system
• changes to data representation
• porting to a new operating system ...

58
Design for Change (Contd)

• Since it is impossible to make everything equally
  easy to change, it is important to
• estimate the probabilities of each type of change
• organize the software so that the items that are
  most likely to change are confined to a small
  amount of code

59
Why is Design for Change Ignored?

• Textbooks fail to discuss the process of
  estimating the probability of change for various
  classes of changes.
• Programmers are impatient because they are too
  eager to get the first version working.
• Designs that result from this principle are
  different from the natural designs of the
  programmers intuition.

60
Why is Design for Change Ignored? (Contd)

• Few good examples of the application of the
  principle. Designers tend to mimic other designs
  they have seen.
• Programmers tend to confuse design principles
  with languages.
• Many practitioners lack training in software
  development.

61
Keeping Records (Documentation)

• Even when software is well designed, it is often
  not documented.
• When documentation is present it is often
• poorly organized
• inconsistent
• incomplete
• written by people who do not understand the system

62
Documentation

• Hence, documentation is ignored by maintainers.
• Worse, documentation is ignored by managers
  because it does not speed up the initial release.

63
Second Opinions (Reviews)

• In engineering, as in medicine, the need for
  reviews by other professionals is never
  questioned.
• In designing a building, ship, aircraft, there is
  always a series of design documents that are
  carefully reviewed by others.

64
Reviews

• This is not true in the software industry
• Many programmers have no professional training in
  software at all.
• Emphasis of CS degrees on mathematics and
  science professional discipline is not a topic
  for a liberal education.
• Difficult to find people who can serve as quality
  reviewers no money to hire outsiders.
• Time pressure misleads designers into thinking
  that they have no time for proper reviews.
• Many programmers resent the idea of being
  reviewed.

65
Reviews

• Every design should be reviewed and approved by
  someone whose responsibilities are for the
  long-term future of the product.

66
Why is Software Aging Inevitable?

• Our ability to design for change depends on our
  ability to predict the future.
• We can do so only approximately and imperfectly.
• Over a period of years
• changes that violate original assumptions will be
  made
• documentation will never be perfect
• reviewers are bound to miss flaws ...

67
Why is Software Aging Inevitable? (Contd)

• Preventive measures are worthwhile but anyone who
  thinks that this will eliminate aging is living
  in a dream world.

68
Software Geriatrics

• Retroactive Documentation
• A major step in slowing the age of older
  software, and often rejuvenating it, is to
  upgrade the quality of the documentation.
• Retroactive Modularization
• Change structure so that each module hides a
  design decision that is likely to change.

69
Software Geriatrics (Contd)

• Amputation
• A section of code has been modified so often, and
  so thoughtlessly, that it is not worth saving.
• Major Surgery (Restructuring)
• Identify and eliminate redundant components and
  gratuitous dependencies.

70
Planning Ahead

• Its time to stop acting as if getting it to
  run was the only thing that matters.
• Designs and changes have to be documented and
  carefully reviewed.
• If its not documented, its not done.
• In other areas of engineering, product
  obsolescence is recognized and included in design
  and marketing plans.
• The same should be done for software engineering.

71
References

• Godfrey96 M. W. Godfrey class lecture notes in
  Software Engineering.
• Parnas86 D. L. Parnas, P. C. Clements, A
  Rational Design Process, How and Why to Fake It,
  IEEE TSE, Vol 12, No 2, 1986.
• Parnas94 D. L. Parnas, Software Aging (plenary
  talk), ICSE, May, 1994.