Contributions in Distributed Systems Engineering

1
Contributions in Distributed Systems Engineering
Thesis Presentation

• by Jari Koistinen

Presenter S. J. Paheerathan
2
Agenda

• Thesis Introduced
• Research Argument
• Claims
• Evidence
• Warrants
• Qualification
• Research Questions
• Research Results
• Authors Contribution
• Thesis Structure

3
Thesis Introduced

• Title Contributions in Distributed Systems
  Engineering
• Author Jari Koistinen
• Type Lit. PhD
• Year 1998
• Pages 412
• Parts V
• Chapters 17

4
Elements of Research Argument

• Claim
• Evidence
• Warrant
• Qualifications

5
Claims, Warrants, Evidences and Qualifications

• Two elements that must be stated explicitly
• Claim
• What you want readers to believe.
• (You should be checked by your doctor)
• Evidence
• Reasons they should believe it.
• (Your blood sugar reading is 200)

6
Claims, Warrants, Evidences and Qualifications

• Warrants
• General principle, an assumption or premise that
  bridges the claim and its supporting evidence
  connecting them into a logically related pair.
• Answers questions not about whether the evidence
  is accurate, but about whether it is relevant to
  the claim or it can be inferred from the evidence
• (Whenever someone has a blood sugar reading of
  more than 120 thats a good sign that she may
  have diabetes)

7
Claims, Warrants, Evidences and Qualifications

• Qualification
• Limits the certainty of conclusions
• Stipulate the conditions in which the claim holds
• Address the readers potential objections
• Make the author to appear a judicious, cautious,
  thoughtful writer

• Your reading is 200evidence, so you should be
  checked,claim because that much glucose in the
  blood is a goodqualification sign that you
  mayqualification have diabetics, warrant unless,
  of course, you just ate something
  sugary.qualification

8
Claims, Warrants, Evidences and Qualifications

• One of the two elements that must be stated
  explicitly
• Claim
• What you want readers to believe.
• (You should be checked by your doctor)

9
Properties of Claims

• They become the heart of the research/report and
  fully reflects the personal contribution of the
  author.
• Are they capable of attracting the readers and
  meet their expectations? (Are they substantive?)
• Will these lead the readers to think?, (Are they
  contestable?)
• Is this something they have long thought? OR
• Is it something they never thought about at all?
• Are they specific/explicit?
• Is it stated in language that is specific?
• Do the readers know what concepts to expect?

10
Claim of the Thesis

• Main Claim
• Improvements are required to the state-of-the-art
  of distributed object systems engineering to
  better meet increasing business and product
  development requirements.

11
Subordinate Claims

• The technologies and computing platforms used to
  support enterprise systems development must have
  at least the following characteristics
• 1. Architecture and implementation techniques
  must enable rapid evolution.
• 2. Systems must satisfy Quality of Service
  requirements such as reliability, security and
  performance requirements.
• 4. Architecture and implementation must enable
  extension and improvement at low cost by
  providing means for reuse.
• 5. Systems must have means for ensuring
  architectural and design integrity as systems
  evolve for many years.

12
Claims, Warrants, Evidences and Qualifications

• One of the two elements that must be stated
  explicitly
• Evidence
• Reasons they should believe it.
• (Your blood sugar reading is 200)

13
Properties of Evidence

• Though the claim is the heart of a thesis, most
  of the thesis will be devoted to supporting the
  evidence
• Do the evidences certain or reliable? (Are they
  accurate?)
• Do the qualifiers set within appropriate limit?
  (Are they precise?)
• Are they sufficient?
• Are they representative?
• Are they authoritative?
• Applicable to the time/current
• Are they perspicuous?
• Can readers see the evidence as evidence?

14
Evidence

• Even recent proposals on high level structuring
  of object oriented systems lack characteristics
  that make them suitable for distributed object
  systems engineering. Classes are too fine-grained
  for cost effective development and reuse of large
  distributed systems.
• The development of large scale distributed object
  systems is more difficult than in necessary as
  common object oriented software engineering
  methodologies do not support separately defined
  and managed interfaces.
• Methodologies today do not support the formal
  description of design invariants, and it is
  difficult to maintain them during long time of
  system evolution, they too do not support
  automatic enforcement of architectural and design
  constraints.

15
Evidence (Cont..)

• QoS aspects such as reliability, security and
  performance are hard to capture and take under
  consideration with commonly used software
  engineering techniques and methodologies.
• Explicit techniques are commonly used only for
  critical systems such as flight control, patient
  monitoring and often limited in dimensions and
  inadequate.

16
Claims, Warrants, Evidences and Qualifications

• Warrants
• General principle, an assumption or premise that
  bridges the claim and its supporting evidence
  connecting them into a logically related pair.
• Answers questions not about whether the evidence
  is accurate, but about whether it is relevant to
  the claim or it can be inferred from the evidence
• (Whenever someone has a blood sugar reading of
  more than 120 thats a good sign that she may
  have diabetes)

17
Properties of Warrants

• They are the basis of our belief and reasoning.
• False warrants
• Unclear warrants
• Inappropriate warrants
• Inapplicable warrants

18
Warrants

• Distributed computing is the infrastructure on
  which present and future enterprises will built
  their information systems.
• enterprises are increasingly becoming global,
  considering the whole world as their market
  place.
• different parts of the business are widely
  distributed and they need to share business
  critical information and functions
• failures or inadequacies in infrastructure will
  be very costly.
• To flourish enterprise must have functionally
  complete, flexible and reliable information
  systems.
• Rapid deployment of new functions and
  incorporation of new sites are critical for
  businesses today.
• Need to accommodate rapid change, search for new
  ways of doing business, technologies.

19
Claims, Warrants, Evidences and Qualifications

• Qualification
• Limits the certainty of conclusions
• Stipulate the conditions in which the claim holds
• Address the readers potential objections
• Make the author to appear a judicious, cautious,
  thoughtful writer

• Your reading is 200evidence, so you should be
  checked,claim because that much glucose in the
  blood is a goodqualification sign that you
  mayqualification have diabetics, warrant unless,
  of course, you just ate something
  sugary.qualification

20
Qualifications

• Development of large scale distributed enterprise
  systems using object orientation.

21
Research Questions

• Is it possible to develop concepts and
  well-defined language for coarse-grained
  modularization of object oriented software such
  that we obtain multi-class components that have
  well defined interfaces and dependencies?
• Is it possible to introduce separately defined
  first class interfaces such that systems are more
  evolve-able and can be naturally mapped to
  distributed object infrastructures without
  imposing any conceptual burdens on developers?
• Whether separating inheritance and
  non-inheritance boundaries, annotating interfaces
  with architectural categories, and aggregating
  sets of tightly-knit interfaces improve the
  internal quality of a software system?

22
Research Questions (Contd..)

• Is it possible to formalise architectural and
  design constraints in a manner that is intuitive
  to use, expressive enough and enables the
  implementation of efficient checking procedures?
• Is it possible to introduce a specification
  technique that enable us to describe, for
  distributed objects, quality of service
  requirements and characteristics involving
  multiple QoS requirements and characteristics
  involving multiple QoS dimensions?
• If so, will it possible to use this technique
  both during the design of multi-class components
  and as a run-time representation for QoS
  information?

23
Research Results

• Concepts, language and tools for modularization
  and structuring of multi-class components and
  interfaces of large object oriented systems.
• Concepts have been implemented in a prototype
  tool and a variation of concepts have been
  integrated with an industrial design language and
  tool environment.
• Concepts have been applied to the design of a
  real telecommunication application and resulted
  in more well-defined architectures and high-level
  structures.

24
Research Results (Contd..)

• Concepts, language and tools for the description
  and enforcement of architectural styles for a
  variety of architectural building blocks such as
  mutli-class components, interfaces and individual
  classes.
• Concepts and language have been given a formal
  semantics based on tree logic. Integration of a
  tree logic inference engine within an industrial
  design environment has been prototyped.
• User evaluation proved that commonly used design
  constraints can be expressed in the language and
  that the formal basis enables efficient
  implementation of real-time checking.

25
Research Results (Contd..)

• Concepts, language and tools for the description
  and exchange of QoS information for object
  oriented components.
• The language has been implemented and a
  corresponding run-time representation and wire
  representation has been defined and implemented.
  It has been used to design a QoS negotiation
  mechanism indicating adequate expressiveness.
• Concepts and language have been used in case
  studies with good results.

26
Contributions of the thesis

• Efficient large-grained structuring of modules
  and interfaces.
• Novel and improved concepts and a language for
  the description and structuring of large grained
  modules and interfaces.
• A refined interface concept separating different
  types of interfaces and grouping of closely
  related interfaces
• case studies and practical experience indicate
  that systems can be built with these concepts to
  be more easily evolved, enabling iterative and
  parallel development, and mapped to distributed
  object technologies.
• The concept and language enables the treatment of
  interfaces as a first class concept in system
  architectures.

27
Contributions of the thesis (Contd..)

• Formal Architectural styles.
• Novel concepts and a language for the formal
  description of invariants for architectural
  designs and design constraints.
• Allows general design languages to be used to
  develop systems with specific architectural
  constraints and characteristics.
• Allows formal treatment and efficient checking
  while providing adequate expressive power.

28
Contributions of the thesis (Contd..)

• QoS specification language and run-time
  representations.
• Novel concepts for QoS specification of object
  oriented components which are separate from any
  syntactic and semantic descriptions of
  components.
• The language balances the expressibility with the
  need for conformance checking of independent
  specifications of a particular interface.
• A well defined language to enable independent
  implementation of infrastructure components that
  can interchange QoS specifications.

29
Structure of the thesis

• Collection of licentiate thesis, four conference
  papers and two research reports.
• Part I - Summarises the research area, research
  questions and results
• Part II - Licentiate Thesis, proposes concepts
  for dealing with programming in the large and a
  case study on a real large scale distributed
  system development
• Part III -Two papers,
• one is on the industrial application of the
  concepts presented in the Lit. thesis and
• other is on further development, formalisation
  and implementation of the architectural style
  concepts mentioned in the Lit. Thesis.

30
Structure of the thesis (Contd..)

• Part IV - Papers and research reports on
  declarative QoS specification on Object Oriented
  components.
• Part V - Epilogue, Concluding remarks.
• Bibliography

31
Thank You for your patience