Back-End Forwarding Scheme in Server Load Balancing using Client Virtualization

Abstract

Load balancing can be done at two levels, the HTTP level and the network level. It refers to the distribution of the HTTP request made by the user across the web server in a server cluster and is achieved by various algorithms such as Random Selection, Round Robin, weighted Round Robin etc. At the network level load balancers are used for the purpose of balancing the load. The approach such as DNS forwarding and spraying the load across the servers is very useful and effective at the network level (11)(12). The load balancers receives the request and proxies the destination ip for forwarding the request to the less utilized servers in order to gain better performance. This paper proposes a server load balancing scheme mainly for the webservers by making the limited number of clients to work as a virtual server which are called virtual clients in the webserver network. This approach is done at the HTTP level by forwarding the HTTP request to the nearest and less loaded virtual client. The HTTP request is only forwarded to get the static content in order to save the current bandwidth utilization of the web server and process more dynamic request on the server side. The daemon routine controls the whole process of web server load balancing but it only processes the HTTP requests. The daemon routine only resides on the web server backend. The client virtualization provides a new anti-overloading technique to reduce the access time of the server and to utilize the server resources without any additional cost. The Client Virtualization also provides a better way to utilize the system bandwidth in web server network. clients the web server easily gets overload thus, it's better to migrate (10) (distribute) the load on other available servers too. The server load also has to be distributed so as not to overload the other available servers in the web-server network. The network congestion in system can reason the overloading which prompts break-down of servers. The best possible appropriation of load on the server is carried out by the load balancer. The load balancer is answerable for the equally assigning the requests to the server pool in order to disallow the over-loading of the single server in the web- server system. The load balancing techniques could be exceptionally adequate for the server hubs to process the client's demands. The load balancing strategies helps in minimizing the reaction time and to legitimately use the throughput of the server network. A large number of algorithms are being utilized by the load balancers to verify which server to send the request to. Random choice, round robin, saturation load balancing, load based, least connection and failover load balancing are the algorithms which are being very frequently used by the load balancers to distribute the load. To enhance the load adjust around the servers various parameters, for example server's recent load, recent reaction time and how much traffic it has been recently assigned are taken into account.