Grid Resource Allocation - Grid Computing: Infrastructure, Service, and Applications - page 169

inversion but also for uncertainty assessment, optimization, control,
and so on. However, the EnKF method is processing intensive due to
iterative simulations of a large number of subsurface models. Large
datasets (comprising model and state vectors) must be transferred from
ensemble-specii c simulation processes to Kalman gain or uncertainty
assessment processes, and back again to the ensemble process at each
assimilation iteration.
Computing synchronization in a grid is used to synchronize all the
components of each iteration of inverse modeling and then initiate the
next iteration. Without effective computing synchronization, application
execution reliability would be seriously impacted by the slowest comput-
ing resource(s) due to the nature of grid resources; that is, geographical
distribution, heterogeneity, and self-administration.

7. 3
Generally speaking, grid resource allocation can be broken into three
stages: resource discovery, resource selection, and job submission [10]. We
will now discuss these separately, but they are inherently intertwined, as
the decision process depends on available information.

Resource Allocation Stages

7. 3 .1 Phase 1: Resource Discovery

Resource discovery involves determining which resources are available to
a given user. At the beginning of this phase, the potential set of resources
is the empty set; at the end of this phase, the potential set of resources is
some set that has passed a minimal feasibility requirement. The resource
discovery phase is done in three steps: authorization i ltering, job require-
ment dei nition, and i ltering to meet the minimal job requirements.
The i rst step of resource discovery for grid scheduling is to determine
the set of resources that the user submitting the job has access to.
Computing over the grid is no different from remotely submitting a job to
a single site: without authorization to run on a resource, the job will not
r u n. At t he e nd of t h i s ste p t he u s er w i l l h ave a l i st of m ac h i ne s or re s ou rce s
to which he or she has access. In the second step, to proceed with resource
discovery, the user must be able to specify some minimal set of job require-
ments in order to further i lter the set of feasible resources. Given a set
of resources to which a user has access and at least a small set of job
requirements, the third step in the resource discovery phase is to i lter out
the resources that do not meet the minimal job requirements. At the end
of this step, the user acting as a grid scheduler will have a reduced set of
resources to investigate in more detail.