Software Architecture A Roadmap

1
Software Architecture - A Roadmap

• Author - David Garlan, CMU
• Presented By - Apurva Jain

2
Agenda

• Roles of Software Architecture
• - The 6 aspects
• The past
• - The 2 trends
• The present
• - ADLs tools, product lines standards,
  codification dissemination
• Supposed future
• - Build vs. buy, network-centric computing,
  pervasive computing
• Conclusion

3
Where does SW-Eng belong?
4
Roles of s/w architecture

• Describes a systems gross structure
• that illuminates top level design decisions
• illustrates the system composition, ways of
  interaction and the properties of system parts
• helps in high level analysis, critical appraisal

5
The six aspects

• System understanding
• Abstraction
• Visibility of constraints and underlying
  architectural choices
• Reuse
• Reuse of component libraries and frameworks
• Construction
• - Acts as a blue prints, highlights components
  and their dependencies (Phillipes 41 model)

6
Phillipes 41 view model
7
The six aspects

• Evolution
• Identifies the load-bearing walls
• Separation of concerns help in evolution
• Analysis
• Enables consistency checking, conformance to
  constraints and QoS, dependence analysis
• Management
• - Helps in understanding requirements, choice of
  implementation strategies and identifying
  potential risks

8
Antecedents of s/w architecture
Courtesy David Garlan
9
The past

• An ad-hoc affair
• Informal box-and-line diagrams
• What the architecture giveth, implementation can
  taketh away
• Diverse, uncodified use of architectural patterns
  and styles
• Architects had to learn it the hard way
• Two trends which highlighted the importance of
  s/w architecture

10
The past the two trends

• Shared set of methods, techniques, patterns etc.
• Use of metaphors such as pipeline, blackboard.
• Significant semantic content which reasoned about
  their paradigms, evolution, relationships and
  concerns
• Exploiting commonalities within specific domains
  to reuse frameworks
• - Some examples are 4GLs, UI toolkits

11
The present

• Recognition of s/w architecture as a sub-field
• - Organizations now have software architects who
  chair the architecture review boards
• Example Bill Gates, CSA, Microsoft Corp.
• - Processes that require architectural design
  reviews and explicit architectural documentation
• - Emerging use of product line architectures,
  commercial architectural standards, component
  integration frameworks
• - Codification of vocabulary, notations tools
  for architectural design
• - Books/courses on software architecture

12
Three significant advancements

• ADLs
• - Need Formal analysis for consistency,
  completeness or correctness, enforcing
  architectural constrains with system evolution
• ADLs provide a framework and a syntax for
  characterizing s/w architectures
• Examples C2, Meta-H, Aesop, Wright etc.
• Different ADLs provide distinct capabilities
  researchers are also working on an architecture
  interchange language called ACME

13
Three significant advancements (continued)

• Product Lines Standards
• Exploiting commonality across products
• Creating product lines within an organization
• Cross-vendor integration standards

14
Three significant advancements (continued)

• Codification Dissemination
• Emergence of architectural styles
• Styles provide a design vocabulary and the
  assumptions and constraints on the use of the
  vocabulary
• Styles give software architects a good starting
  point
• Example If its a data flow application, my
  intuition will suggest me that I use a pipe and
  filter style.

15
The future

• Current trends have a daunting effect on the way
  software architecture will shape up for tomorrow
• - Build vs. Buy
• Time to market has become an important criteria
  in decision-making hence many tend to buy COTS
• Integrating in-house components with vendors
  components is a nightmare
• Most engineers nowadays only write glue-code
• Need for industry wide standards such as EJB, HLA

16
The future (continued)

• - Network-centric computing
• Institutions and their respective designers do
  not have control over the system, knowledge of
  the operating environment, and availability of
  resources and services which are sporadic
• Such systems demand new architectures which can
  scale up to the size and variability to that of
  the internet for example
• Challenge is to manage dynamically evolving
  components which need to analyzed and managed at
  run-time
• Need for architectures which give the flexibility
  to end users to compose their own systems with
  some guarantee on integration and overall
  performance.

17
The future (continued)

• Pervasive computing
• Poses a challenge to identify architectures that
  allows modifying the fidelity of computation
  based on power reserves
• The pervasive environment is one where devices
  come and go and therefore the need to dynamically
  reconfigure in real time.
• Theres also the need for architectures that will
  support user mobility in the face of changing
  environments.

18
Conclusion

• Good analysis of s/w architecture timeline
• Speculated future work then is showing up in
  todays conferences
• Concerns raised for developing pervasive systems
  overlap with what we have been discussing in
  class
• Devils normally reside in the details and this is
  a roadmap paper pretty abstract