Package org.palladiosimulator.pcm.seff

Interface LoopAction

All Superinterfaces:
AbstractAction, AbstractInternalControlFlowAction, AbstractLoopAction, org.eclipse.emf.cdo.CDOObject, org.eclipse.emf.cdo.common.id.CDOWithID, Entity, EObject, de.uka.ipd.sdq.identifier.Identifier, NamedElement, Notifier, PCMBaseClass, PCMClass
All Known Implementing Classes:
LoopActionImpl
public interface LoopAction extends AbstractLoopAction
A representation of the model object 'Loop Action'. Models the repeated execution of its inner ResourceDemandingBehaviour for the loop body. The number of repetitions is specified by a random variable evaluating to integer or an IntPMF. The number of iterations specified by the random variable always needs to be bounded, i.e., the probabilities in an IntPMF for iteration numbers above a certain threshold must be zero. Otherwise, it would be possible that certain requests do not terminate, which would complicate performance analyses. The stochastic expression defining the iteration random variable may include references to input or component parameters to model dependencies between the usage profile and the number of loop iterations. Notice, that loop actions should only be modelled if the loop body contains either external service calls or resource demands directed at special resources. Otherwise, control flow loops in component behaviour should be abstracted by subsuming them in InternalAction, which combine a number of instructions. The influence of different iterations length of such internal loops need to be reflected stochastically by the random variable specifying the ParametricResource-Demand of that InternalAction. Other than Markov chains, RDSEFFs do not specify control flow loops with an reentrance and exit probability on each iteration. Such a specification binds the number of loop iterations to a geometrical distribution, which reflects reality only in very seldom cases. But in many practical cases, the number of iterations is a constant, or the probability for higher iteration numbers is higher than for lower ones. This cannot be expressed directly via a Markov chain (also see [DG00]). Inside the ResourceDemandingBehaviour of LoopActions, it is assumed that random variables are stochastically independent. This is not true in reality, and for example leads to wrong predictions if the same random variable is used twice in succession inside a loop body. In this case, the second occurrence is stochastically dependent to the first occurrence, as the value does not change between two occurrences. Therefore, component developers should be aware of such inaccuracies when using random variables twice inside the body behaviour of a LoopAction.

The following features are supported:

See Also:
Generated class or method.
EMF model class or method.