Information Technology Reference
In-Depth Information
engine is the main component of the GRUBER architecture and that implements
various algorithms to detect available resources and maintains a generic view of
resource utilization in the grid [5]. The proposed mechanism allows resources at
individual sites to be shared among multiple user communities. It is a distributed
Grid brokering service that it addresses issues regarding how USLAs can be stored,
retrieved, and disseminated efficiently in a large distributed environment. DI-
GRUBER has some defect in resource matching to tasks and repartition [17], but
other recommendations have been offered to solve its problem.
The eNANOS Resource Broker is an OGSI-Compliant resource broker developed
as a Grid Service and is supported by Globus Toolkit middleware [6]. This Broker
provides a set of Grid Service interfaces and a Java API which can be used from
command-line clients, applications or portals. The eNANOS uses from the Grid In-
formation Service (GIS) to obtain needed information. As you know, GIS has some
problems such as data restriction, the timeworn data, updateability problem and so
forth. eNANOS architecture also neither uses data mining methods to select the best
nodes from the pool of discovered nodes, nor implements in Web Services (WS)
bases frameworks.
The AppLes (Application Level Scheduling) focuses on developing scheduling
agents for individual Grid applications [7]. AppLes agents have an application ori-
ented scheduling mechanism, and use static or dynamic application and resource
information to select a set of resources. However, they perform resource discovering
and scheduling without considering resource owner policies. Also they do not support
system-oriented or extensible scheduling policies.
Another resource broker service has been presented by Young-Seok Kim and et al.
[4]. It is an OGSI- based broker that is supported by GT3. It is a new general purpose
OGSI-compliant Grid resource broker service that performs resource discovering and
scheduling with close interactions with GT3 Core and Base Services. This resource
broker service considers resource owner policies as well as user requirements on the
resources.
The EZ-Grid project [8] tries to develop a transparent view of Grid resources and a
simplified job submission through the EZ-Grid resource broker. The system is devel-
oped in Java, using the Java CoG Kit. The aim of the system is to relieve the user
from the complexity involved in making resource selections, job specifications and
submissions while maintaining transparency in providing middleware services and
best resource choices. The system provides an easy-to-use interface that allows for
user authentication, application profiling, remote resource information display, job
submission and job monitoring.
Another works have been done in resource selection field (e.g. Condor/G [15],
Nimrod/G [18], and so forth), but we cannot introduce all of them in this paper.
Finally, we mention that none of those systems or brokers uses machine learning
methods to find (select) the best nodes for purposed jobs.
3 Fuzzy Decision Tree
Decision Tree (DT) is one of the most popular methods for learning and reasoning
from feature-based examples. Due to following causes DT is better for our work:
Search WWH ::
Custom Search