of this report is to show how the round robin and weighted round robin are
developed for solving network based traffic problems. Nowadays Application for
the web are developing violently, causing
gigantic of system loads. This high development rate
leads to decrease in performance and responding time in networks. The solution
to resolve this kind of traffic issues is using Load balancer with scheduling
algorithm between the user requesting the job and the server. It creates a
multiple server distributed system to distribute the loads. Fig.1 shows how the
network would look like from an overview. Since an arrangement with distributed
loads around its multiple server’s yields in better Network performance.
many different scheduling algorithm for load balancer. Mainly the Scheduling
algorithm is divided into two types. The first one being preemptive and the second
one being non-preemptive. Preemptive scheduling can be interrupted during the
process but the non-preemptive can’t be interrupted between processes. There a
lot of criteria looked to compare a good scheduling algorithm (www.tutorialspoint.com, 2018):
CPU Utilization: Algorithm should always keep the CPU be
busy as much as possible. This makes the use of CPU more efficient algorithm
and better network.
Throughput: It’s the number of jobs possible to be done
by system in an hour. It should be optimised to do Maximum number of jobs per
hour for efficient algorithm.
Response time: The time taken for the Load balancer to
start responding to the request. There should be less response time for efficient
Turnaround time: The time user has to wait for the job
request to be completed. The time should be minimum for efficient algorithm.
Waiting time: It’s the time a job waits to be assigned to
the node to complete its work. This should also be minimised for efficient al
Fairness: A good scheduler make sure that all process get
fair share of the CPU.
mainly focuses on two scheduling algorithm. Initially using the round robin and
then weighted round robin to control the load more efficiently in distributed