Abstract
A typical problem when executing simulation of large and complex
Coloured Petri Nets (CPN) is the excessive demand for computational
resources. In some cases, slowness is the major problem, but in extreme
situations, simulation can abort before completion. In addition, to
analyze and extract results from large simulation models can be quite
complex. An alternative way to deal with these problems is to fragment
large CPN models into smaller nets and conceive models as distributed
systems (DS). In this work, we present the Distributed Fusion Places
(DFP), one new node to compose distributed CPN, which support
communication and interaction between geographically dispersed subnets,
which together make up a larger, unique and coherent model. In this way,
different processes execute subparts of a single model, processing part
of the computational effort needed to simulate the global model. Results
obtained from distinct parts of the model can be analyzed individually
or consolidated to represent the overall system operation. To
demonstrate and validate this approach, we have introduced CacheSIM, a
simulator conceived to analyze file admission and substitution policies
in systems with distributed caches.
Users
Please
log in to take part in the discussion (add own reviews or comments).