Databases Reference
In-Depth Information
Apart from the systems point of view, cloud computing also has its origin
from the Internet infrastructural perspective. The cloud computing capability
is derived from the advancement of Internet and web technologies. In addi-
tion, cloud computing has benefited from the rapid development of communi-
cation technologies. Through large-scale network deployments, such as a grid,
wide-area networks (WAN), and the Internet, users are able to do large-scale
computations using either dedicated or shared network resources.
6.1.2.1.1 Definitions
The term grid computing was coined in the mid-
1990s to describe technologies that allow consumers to obtain computing re-
sources on demand. These services were initially intended for advanced sci-
ence and engineering applications. A well-structured definition of cloud com-
puting was introduced by Foster et al. [76]: “A large-scale distributed com-
puting paradigm that is driven by economies of scale, in which a pool of ab-
stracted, virtualized, dynamically scalable, managed computing power, storage,
platforms, and services are delivered on demand to external customers over
the Internet.”
Cloud computing can be differentiated from other existing distributed sys-
tems in the following ways:
• Cloud computing is massively scalable. Resources can be instantaneously
added and/or removed.
• Cloud computing is driven by economies of scale.
• Cloud computing can deliver different levels of services to customers
and offers abstraction of implementation, i.e., users are unaware of the
details of the underlying application deployment.
6.1.2.1.2 Cloud addresses grid limitations
Heterogeneity is an im-
portant issue in grid computing implementations. Large sets of heterogeneous
hardware and software owned by different organizations with different usage
policies pose a challenge to grid-based applications. Middleware applications
and libraries on top of the grid fabric offer a mechanism for homogenization.
However, di
culties still arise in cross-grid applications that run across differ-
ent numbers of virtual organizations (VOs). In normal configurations, users
are likely to be entangled in complex resource management and facilitation
tasks.
In contrast, cloud systems emerged to provide a determined set of capabil-
ities to the users; interface is the purpose of the design. These systems are
often developed using a top-down design, in which the usability of the inter-
face, i.e., the application, is the primary goal. Furthermore, cloud systems can
be developed on top of existing grid infrastructures. Table 6.1 shows some of
the differences between cloud and grid systems, as described by Foster et al.
[76].
