Summary: | Load Balancing is the solution to a problem of the redundancy in the
network. Redundancy of itself in the context of the network has a network of
terms, i.e., the ability to survive in a density of traffic, where the network is able
to survive to continue serving packages of input and output, providing alternative
routes when there is an error or a path is down. Some studies propose alternative
solutions to the problem through the construction of a computer network through
the addition of new servers when one increases the redundancy with the number
of users has increased. But the solution isn't to not cause new problems. The
problem then is an unequal burden-sharing on the routes that are used to handle
the load of traffic. Some recent research gathered at the conclusion that the best
method of load balancing for dedicated load-balancing method using the softwarehardware.
This study compares the two methods to evaluate the traffic, the
composition of the protocols, switching by error, the number of packages,
channels of distribution, performance, TTL, the packet loss rate, mangle settings,
and system resources. All parameters (Quality of Service) QoS are compared
using the standard deviation. Recommendations based on the smallest primary
school are, and the calculation of the analysis of log files that can determine the
capacity of the method of load balancing to distribute the load. Observations of
the traffic and the quality of services that are produced in this intense search tools
for monitoring network, such as cactigraph, Winbox and ntop.
The results showed that the CLN method has standard deviation values
smaller than the CCP in terms of quality of resulting service. With regard to the
behaviour of the distribution of the burden, the method n distribute the burden
more evenly to the line's existing servers, while the CCP algorithm method tends
to maximize the resources of a server.
|