Information Technology Reference
In-Depth Information
18.1
Simulation is a low-cost and safe alternative for solving complex problems
in various areas such as production, business, education, science, and
engineering. With increased problem size and the advance of parallel
computing, parallel simulation was introduced to increase the simula-
tion speed. With the worldwide prevalence of networked computers, dis-
tributed simulation was introduced to promote the interoperability and
reu s abi l it y of si mu l at ion appl ic at ion s a nd to l i n k ge og raph ic a l ly d i sper s e d
simulation components. The high-level architecture (HLA) has been stan-
dardized as IEEE 1516 for distributed simulation in September 2000 [1].
While HLA dei nes the rules, interface specii cation and object model
template (OMT), a runtime infrastructure (RTI), such as the Defense
Modelling and Simulation Ofi ce (DMSO) RTI [2], provides the actual
implementation of the HLA standard. In HLA terminology, a distributed
simulation application is called a federation that comprises several
simulation components called federates; the RTI is a communication mid-
dleware through which federates in the same federation can communi-
cate with each other.
Traditionally, HLA-based distributed simulations are conducted using
a vendor-specii c RTI software and federates with different RTI versions
cannot cooperate with each other. To run a distributed simulation over a
WAN, the required software and hardware resource arrangements and
security settings must be made before the actual simulation execution.
Because of this inl exibility, it is not easy to run HLA-based distributed
simulations across administrative domains. To address these inl exibility
issues and leverage globally pervasive resources for distributed simula-
tions, the grid is naturally considered as a solution.
Grid computing was proposed by Foster as a l exible, secure, and coor-
dinated resource shared among dynamic collections of individuals, insti-
tutions, and resources in different virtual organizations [3]. Among the
various available grid middlewares, Globus Toolkit [4] is the
de facto
stan-
dard middleware for grid computing. Based on open grid services archi-
tecture (OGSA) [5], its latest version GT4 contains i ve components, namely
common runtime, security, data management, information services, and
execution management, to facilitate heterogeneous resource sharing.
Merged with Web service standards, GT4 implements a group of related
Introduction
Search WWH ::
Custom Search