Database Reference
In-Depth Information
The growth and development of the IT industry in the twenty-first century is pri-
marily driven by the need to provide more and more domain-specific flexible and
innovative functionality at more and more commoditized, low-cost, distributed,
multilayered, heterogeneous, and interoperable architectural infrastructures.
10.1 Distributed Applications
Distributed applications are constituted of a collection of heterogeneous but fully autonomous
components that can execute on different computers. While each of these components has full
control over their constituent subparts, there is no master component that possesses control over
all the components of a distributed system. Thus, for the system to appear as a single and inte-
grated whole, the various components need to be able to interact with each other via predefined
interfaces through a computer network.
The characteristic global features of a successful distributed application are as follows:
1.
Distributed systems are heterogeneous
arising from the need to (say) integrate components on a
legacy IBM mainframe, with the components newly created to operate on a UNIX worksta-
tion or Windows NT machine.
2.
Distributed systems are scalable
in that when a component becomes overloaded with too many
requests or users, another replica of the same component can be instantiated and added to
the distributed system to share the load among them. Moreover, these instantiated compo-
nents can be located closer to the local users and other interacting components to improve
the performance of the overall distributed system.
3.
Distributed systems execute components concurrently
in a multithreaded mode via multiple
invoked components corresponding to the number of simultaneously invoked processes.
4.
Distributed systems are fault tolerant
in that they duplicate components on different comput-
ers as that if one computer fails, another can take over without affecting the availability of
the overall system.
5.
Distributed systems are more resilient
in that whereas distributed systems have multiple
points of failure, the unaffected components are fully operational even though some of
the components are not functional or are malfunctioning; moreover, the distributed sys-
tem could invoke another instance of the failed components along with the corresponding
state of the process (characterized by the program counter of the process, the register vari-
able contents, and the state of the virtual memory used by the process) to continue with
the process.
6.
Distributed systems demonstrate invariance or transparency
with reference to characteristics
like
a. Access locally or across networks to the components
b. Physical location of the components
c. Migration of components from one host to another
d. Replication of components including their states
e. Concurrency of components requesting services from shared components
f. Scalability in terms of the actual number of requests or users at any instance
g. Performance in terms of the number and type of available resources
h. Points of failure, be it a failure of the component, network, or response
