A tutorial: Load balancers in a container technology system using docker swarms on a single board computer cluster

Abstract

Overload requests on a webserver are one of the important issues in web server management.
The main cause of overload requests on the webserver is the high demand by the user / client for the
webserver that exceeds the maximum load limit borne by the webserver. One of the ways used to handle
overload requests on the webserver is by sharing the requests evenly among the available web servers.
The webserver management system still uses a single server by the webserver management developer
which results in the possibility of overloading requests on the webserver. The method proposed in this
study is a haproxy load balancer on a single board computer cluster using a docker swarm. Before the
performance testing analysis stage is carried out, the researcher will build a single board computer
cluster, using 1 raspberry pi master server which will become the load balancer and 3 raspberry pi
worker servers to become a webserver which will respond to users alternately according to the algorithm
that will be used on the hap Roxy load balancer. After testing in this study, it was concluded that using the
load balancing method with the least connection algorithm resulted in an accuracy rate of 95.47%, while
testing using the load balancing method with the round robin algorithm was able to produce a greater
level of accuracy, namely 97%. Use 1 raspberry pi master server which will be the load balancer and 3
raspberry pi worker servers to become the webserver which will respond to the user alternately
according to the algorithm that will be used on the haproxy load balancer. After testing in this study, it
was concluded that using the load balancing method with the least connection algorithm resulted in an
accuracy rate of 95.47%, while testing using the load balancing method with the round robin algorithm
was able to produce a greater level of accuracy, namely 97%. use 1 raspberry pi master server which will
be the load balancer and 3 raspberry pi worker servers to become the webserver which will respond to
the user alternately according to the algorithm that will be used on the haproxy load balancer. After
testing in this study, it was concluded that using the load balancing method with the least connection
algorithm resulted in an accuracy rate of 95.47%, while testing using the load balancing method with the
round robin algorithm was able to produce a greater level of accuracy, namely 97%.