SABLE: Agent Support for the Consolidation of Enterprise-Wide Data-Oriented Simulations
Simulations are programs that model some real world situation and mirror
their interactions in the software domain. In most cases, conclusions
can be made based on the analysis of initial and resulting data of the
simulation. Consequently, large amounts of data are processed and stored,
as both inputs and outputs of the simulation. Handling data is common
to most simulation domains, however there is limited specialized support
for simulation-based data management. Most simulation environments utilize
batch files (large text files) to both supply initial data and record
resultant data. This solution is quick and accessible, but there are
several problems. One problem is standardization.
Even within small corporations there can be multiple data formats for
batch files, reducing interoperability between simulation programs.
A second problem is the tightly coupled dependency between the correct
operation of the simulation and consistency of the file formats and
file directory structure. Moreover within one corporation, there can
be redundant implementations of file streaming methods. Developers also
neglect the benefits of persistence and concurrency that can be provided
by a Database Management System (DBMS).
This paper introduces SABLE (Simulation-oriented Agent-Based Library
and Environment) as a solution to these problems. SABLE is an agent-controlled
environment that enforces the policy of a specialized data format, integrates
a DBMS, and removes the task of data management from the developers
of the simulation program.
