Classical Architectural Styles

Classical Architectural Styles

 Common architectural idioms, taxonomies,
and patterns
 Issues
 Detailed look at specific architectural styles
 Pure forms first, later heterogeneous systems
 Distinguishing characteristics and specialisation
 Heuristics for choosing a style
 Implementation techniques
 Formal models and analysis
 Case studies
Subtopics
 Dataflow systems
 Batch sequential, pipe & filter
 Procedure call systems
 Information hiding, ADTs, objects
 Event-based systems
 Multicast organisation, implicit invocation
Subtopics con’t
 Repository-oriented systems
 Blackboards, databases, client-servers
 Processes
 Communicating processes, message
passing
 Others
 Interpreters and other virtual machines,
process control, layers
Architecture in Systems
 Architecture:
“the underlying structure of things”
Not just “what”, but “why”
 Good architecture (like much good design):
 Result of a consistent set of principles and
techniques, applied consistently through all
phases of a project
 Resilient in the face of (inevitable) changes
 Source of guidance throughout the product lifetime
Elements of Architectural
Descriptions
 The architectural definition of a system
selects
 Components: define the locus of computation
 Examples: filters, databases, objects, ADTs
 Connectors: mediate interactions of components
 Examples: procedure calls, pipes, event broadcast
 Properties: specify info for construction & analysis
 Examples: signatures, pre/post conds, RT specs
Elements of Architectural
Descriptions con’t
 An architectural style defines a family of
architectures constrained by
 Component/connector vocabulary
 Topology rules
 Semantic constraints
Common Architectural Idioms
 Data flow systems
Batch sequential Pipes and filters
 Call-and-return systems
Main program & subroutines
Hierarchical layers OO systems
 Virtual machines
Interpreters Rule-based systems
 Independent components
Communicating processes Event systems
 Data-centered systems (repositories)
Databases Blackboards
Batch Sequential Systems
 Processing steps are independent programs
 Each step runs to completion before next step
starts
 Data transmitted as a whole between steps
 Typical applications
 Classical data processing
 Program developments
Pipes and Filters
 Filter
 Incrementally transform some amount of
the data at inputs to data at outputs
 Stream-to-stream transformations
 Use little local context in processing stream
 Preserve no state between instantiations
Pipes and Filters con’t
 Pipe
 Move data from a filter output to a filter
input
 Pipes form data transmission graphs
 Overall computation
 Run pipes and filters (non-deterministically)
until no more computations are possible
Main Program and
Subroutines
 Hierarchical decomposition
 Based on definition-use relationship
 Single thread of control
 Supported directly by programming languages
 Subsystem structure implicit
 Subroutines typically aggregated to modules
 Hierarchical reasoning
 Correctness of a subroutine depends on the
correctness of the subroutines it calls
Object architectures
 Encapsulation
 Restrict access to certain information
 Inheritance
 Share one definition of shared functionality
 Dynamic binding
 Determine actual operation to call at runtime
 Management of many objects
 Provide structure on large set of definitions
 Reuse and maintenance
 Exploit encapsulation and locality
Layered Patterns
 Each layer provides certain facilities
 Hides part of lower layer
 Provides well-defined interfaces
 Serves various functions
 Kernels: provide core capability, often set of
procedures
 Shells, virtual machines: support for portability
 Various scoping regimes
 Opaque versus translucent layers
Interpreters
 Execution engine simulated in software
 Data
 Representation of program being interpreted
 Data (program state) of program being interpreted
 Internal state of interpreter
 Control resides in “execution cycle” of
interpreter
 But simulated control flow in interpreted program
resides in internal interpreter state
 Syntax-driven design
Communicating Processes
 Components: independent processes
 Typically implemented as separate tasks
 Connectors: message passing
 Point-to-point
 Asynchronous and synchronous
 RPC and other protocols can be layered on
top
Event Systems
 Components: objects or processes
 Interface defines a set of incoming procedure calls
 Interface also defines a set of outgoing events
 Connectors: event-procedure bindings
 Procedures are registered with events
 Components communicate by announcing events
at “appropriate” times
 When an event is announce the associated
procedures are (implicitly invoked)
 Order of invocation is non-deterministic
 In some treatments connectors are event-event
bindings
Classical Databases
 Central data repository
 Schemas designed specifically for application
 Independent operators
 Operations on database implemented
independently, one per transaction type
 Interact with database by queries and updates
 Control
 Transaction stream drives operation
 Operations selected on basis of transaction type
The Blackboard Model
 Knowledge sources
 World and domain knowledge partitioned into
separate, independent computations
 Respond to changes in blackboard
 Blackboard data structure
 Entire state of problem solution
 Hierarchical, nonhomogeneous
 Only means by which knowledge sources interact
to yield solutions
 Control
 In model, knowledge sources self-activating
Important ideas
 Common patterns for system structure
 Pure type form, allowing variation
 Identifiable types of subsystems and interaction
 Decomposition and heterogeneity
 Pattern also describe subsystem structure
 Subsystem pattern o system pattern
 Independence
 System patterns and subsystem functions do not
depend on application
 Fit to problem
 Problem characteristics guide choice of structure
Case studies: 1. KWIC
 In his paper of 1972 Parnas proposed the
following problem [Parnas72]
The KWIC (Key Word in Context) index system
accepts an ordered set of lines; each line is an
ordered set of words, and each word is an ordered
set of characters. Any line may be “circularly
shifted” by repeatedly removing the first word and
appending it at the end of the line. The KWIC
index system outputs a listing of all circular shifts
of all lines in alphabetical order.
KWIC con’t
 Four solutions:
 Shared data
 ADT
 Implicit invocation
 Pipe-and-filter
KWIC con’t
 Consider following changes:
 Changes in the processing algorithm
 Changes in the data representation
 Enhancement to system function
 Performance
 Reuse
2. Instrumentation software
 Develop a reusable system architecture for
oscilloscopes
 Rely on digital technology and
 Have quite complex software
 Reuse across different oscilloscope products
 Tailor a general-purpose instrument to a specific
set of users
 Performance important
 Rapid configuration of software within the
instrument
 => Domain specific software architecture
Instrumentation software con’t
 Object-oriented model of software domain
 Clarified the data types used for oscilloscopes
 Waveforms, signals, measurement, trigger modes, …
 No overall model to explain how the types fit
together
 Confusion about partitioning of functionality
 Should measurements be associated with types of data
being measured or represented externally?
 Which objects should the user interface interact with?
Instrumentation software con’t
 Layered model
 Well-defined grouping of functions
 Wrong model for the application domain
 Layer boundaries conflicted with the needs of
the interaction among functions
 The model suggest user interaction only via
Visualization, but in practice this interaction affects all
layers (setting parameters, etc)
Instrumentation software con’t
 Pipe-and-filter model
 Functions were viewed as incremental
transformers of data
 Corresponds well with the engineers’ view
of signal processing as a dataflow problem
 Main problem:
 How should the user interact?
Instrumentation software con’t
 Modified pipe-and-filter model
 Each filter was associated with a control interface
 Provides a collection of settings to be modified
dynamically by the user
 Explains how the user can make incremental
adjustments to the software
 Decouples signal-processing from user interface
 Signal-processing software and hardware can be
changed without affecting the user interface as
long as the control interface remains the same
Instrumentation software con’t
 Further specialisation
 Pipe-and-filter lead to poor performance
 Problems with internal storage and data
exchange between filters
 Filters may run at radically different speeds
Instrumentation software con’t
 Summary
 Software must be typically adapted from
pure forms to specialised styles (domain
specific)
 Here the result depended on properties of
pipe-and-filter architecture adapted to
satisfy the needs of the product family