Design and Performance Evaluation of a Service-Oriented HLA RTI on the Grid - Grid Computing: Infrastructure, Service, and Applications - page 451

vendor-specii c RTI. This effort also follows the grid-enabled approach.
Recently, Chen et al. [23] integrated the federate-proxy-RTI based frame-
work with the HLA RePast middleware system and therefore proposed an
HLA grid RePast platform for executing large-scale agent-based distrib-
uted simulation on the grid [24]. The major drawback of the grid-enabled
approach is that vendor-specii c RTI execution environments and com-
munication servers have to be set up beforehand, which makes this
approach not so l exible.

18.2.3 Grid-Oriented Approach

In the grid-oriented approach, the RTI is implemented using grid services
according to the HLA specii cation. The six HLA service groups should be
mapped to different grid services in order to create a service-oriented
architecture. This approach was raised in Fox's keynote at the Distributed
Simulation and Real Time Applications Conference in 2005 [25]. Our
SOHR framework follows the grid-oriented approach and its design is
discussed in the next two sections.

18.3
The architecture of our SOHR framework is illustrated in Figure 18.1 . It
contains seven key grid services, namely the RTI index service, the local
service (LS), and i ve management services, all of which are implemented
based on GT4. All services except the RTI index service follow the
WS-Resource factory design pattern [26].
In the WS-Resource factory design pattern, information is organized
into resource instances: a resource home keeps track of the multiple
resource instances: a factory service is dei ned to create new resource
instances; an instance service uses the resource home to i nd the client-
specii ed resource instance and operates on it.
The RTI index service provides a system-level registry so that all other
services are able to register their end point references (EPRs) here and
dynamically discover each other. It also provides services to create and
destroy federations in the system. The i ve management services corre-
spond to the six HLA service groups. Each of them consists of a factory
service, an instance service, and multiple resource instances, with one
resource instance for each federation. For example, the Federation
Management service (FMS) provides functionalities of the HLA Federation
Management service group as a grid service and consists of the factory
service FMFS, the instance service FMIS and multiple resource instances
FMRIs, with one FMRI for each federation.

Framework Overview of SOHR