Classification of the Palladio Component Model (PCM)

The following listing helps to classify the Palladio Component Model (PCM) which is underlying the Palladio approach. In case you are preparing a taxonomy or try to identify whether specific features are supported by the PCM, this page assists your work.

• Supported quality dimensions
  • Performance
  • Reliability
  • Costs
  • Maintainability
• Meta-Meta-Model: UML2 and ECORE representation are available
• Meta-Model: Own meta-model (i.e. no UML profile); the meta-modell posseses execution semantics; OCL constraints ensure model integrity
• Meta-Model implementation: Eclipse EMF and Java
• Interface model: Corba IDL-like
• Supported compositions
  • Vertical: Composition by component assembly
  • Horizontal: Composition by component allocation
  • Hierarchical / nesting: Component encapsulation by composite components, subsystems, and systems
• Component modelling
  • Three different component types (Provided Component Type, Complete Component Type, Implementation Component Type)
  • Two different component realisations (Composite Components, Basic Components)
  • Multiple type and instance levels (assembly, allocation, usage)
• Model partitioning
  • Component repository: Definition of components and interfaces
  • System: Definition of component assembly
  • Execution environment: Definiton of hardware nodes and network
  • Component Allocation: Definition of how component are deployed on hardware nodes
• Supported developer roles (can interact independently)
  • Component developer
  • Software architect
  • Component deployer
  • QoS analyst
• Behaviour Model (describes single provided services of a component)
  • Internal and external actions
  • Sequences
  • Alternatives (probabilistic + guard conditions)
  • Loops (probabilistic + guard conditions)
  • Concurrency support: Fork / join / barrier
  • Semaphores
• Interaction
  • Request / reponse (call / return, synchronous)
  • Event-based communication (asynchronous)
• Parameterized component dimensions (i.e. dimension can vary for a readily defined component type)
  • deployment / allocation
  • usage
  • assembly
  • implementation
• Stochastic Expression language for expressing parameterised control and data flow
  • Constants
  • Distributions
  • Mathematic calculations
  • Conditions
  • Equality and inequality checks
• Supported states
  • Request-based (for each invocation of a provided service of a component)
    • Behaviour model state (which behaviour steps is currently executed)
    • Parameter values and paramter propagation per service request
    • Local variables
  • Allocation-based (component configuration parameters per component instance)
  • During analysis: Contention effects / blocking
  • Further state is automatically derived from allocation, assembly, and usage of components (e.g. how a component is allocated)
  • Interface protocols (mandatory orders for calling component services of an interface): Supported but not implemented

Classification of the Palladio Component Model Analyses Approaches

• Analysis output
  • Performance: General distribution functions
  • Reliability: Mean value
  • Costs: Sum of component costs (e.g. development costs, procurement costs or total cost of ownership) and hardware costs (e.g. procurement costs, operating costs or total cost of ownership)
  • Maintainability: Expected maintainability costs

Special Features

• The PCM accounts for the performance impact of exchangeable middlewares, e.g. application servers (“performance completion”)
• Derivation of service level agreements (SLA) from analysis results

Extended Support

• Java stub generation from the PCM model
• EJB stub generation from the PCM model
• Performance prototype generation (puts load on real hardware as specified in the model)

NOTE: The classification of the Palladio Component Model given in Crnković, I.; Sentilles, S.; Vulgarakis, A.; Chaudron, M. R V, “A Classification Framework for Software Component Models,” Software Engineering, IEEE Transactions on , vol.37, no.5, pp.593-615, Sept.-Oct. 2011, doi: 10.1109/TSE.2010.83 is partially wrong. Please refer to the above classification.