Grid Computing - Guide to Cloud Computing for Business and Technology Managers - page 249

So the key concept emphasized was the ability to negotiate resource sharing
agreements among a set of participating parties—where sharing did not really
mean exchange but direct access to computing resources either in a collabora-
tive resource sharing or negotiated resource-brokering strategies. Further,
this sharing was highly controlled with resource providers and consumers
grouped into virtual organizations primarily based on sharing conditions.
The following is the precise simple checklist that was proposed: a grid is a
system that functions as follows:
1. Coordinates resources that are not subject to centralized control : The first
criterion states that a grid should integrate computing resources
from different control domains (say servers from computer cen-
ters of different universities, each center having a different system
administrator in each university). Technologically, this requirement
addresses the issues of cross-domain security, policy management,
and membership.
2. Uses standard, open, general-purpose protocols and interfaces : The use of
a common standard for authentication, authorization, resource dis-
covery, and resource access becomes a necessity in such cases and
hence the second criterion.
3. Delivers nontrivial quality of service : Finally, in an effort toward com-
mercializing the usage of shared resources, it is important to sup-
port various quality-of-service parameters such as response time,
throughput, availability, or even co-allocation of resources to meet
user demands.
12.5 Types of Grids
Broadly speaking, there are three types of grid: computational grids, data
grids, and service grids. A computational grid is a distributed set of resources
that are dedicated to aggregate computational capacity. Computational grids
are highly suitable for task farming or high-throughput computing appli-
cations where there is typically one data set and a huge parameter space
through which the scientist wishes to search. A data grid is a collection of dis-
tributed resources that are specifically set up for processing and transferring
large amounts of data. A service grid is a collection of distributed resources
that provides a service that cannot possibly be achieved through one single
computer. In this example, therefore, the grid will typically consist of several
different resources, each providing a specific function that needs to be aggre-
gated in order to collectively perform the desired services. For example,
you could have a service that obtained its functionality by integrating and