China Grid and Related Dependability Research - Grid Computing: Infrastructure, Service, and Applications - page 104

handling method without having any knowledge about the grid
resources and the applications.
Task-specii c failure handling support. This is driven by the diver- •

sity of grid applications and grid resources. The user should be
able to specify an appropriate failure handling method for perfor-
mance and cost consideration.

In this section, an adaptive policy-based fault-tolerance approach is
presented, which is called DRIC. First the overview of the approach is
presented, with application-level fault-tolerance techniques reviewed
later. Finally, the adaptive model of failure handling will be presented.

4.5.4.1 Overview of Failure Handling

As depicted in Figure 4.12, the fault tolerance in DRIC framework com-
prises two phases: failure detection and failure handling. Figure 4.12
presents the overview of the failure handling approach. The failure-
handling approach uses a decision-making method to attain the QoS
requirements described by users, in which almost all kinds of application-
level failure-recovery techniques are integrated, such as checkpointing
[59-62], replication [63-66], and workl ow [67]. First, the user submits a
task with QoS requirements. The policy engine analyzes the QoS
requirements and the attributes of the application to constitute a fail-
ure-recovery policy. Based on the policy, the policy engine carries out
the policy with appropriate techniques with the help of job manage-
ment, data management, and information management illustrated in
Figure 4.13 .

Application/user interface

Failure detector Failure handling

Data
collector

Policy
maker

Policy
executor

Index
service

Failure-recovery
techniques

Heartbeat
monitor

Resource management

FIGURE 4.12 Overview of failure handling.