Policies for Advanced Dynamic Load Balancing in Cloud Computing

Policies for Advanced Dynamic Load Balancing in Cloud Computing

Dharmesh Dhabliya, Sukhvinder Singh Dari, Nitin N. Sakhare, Anish Kumar Dhablia, Digvijay Pandey, A Shaji George, A. Shahul Hameed, Pankaj Dadheech
Copyright: © 2024 |Pages: 14
DOI: 10.4018/979-8-3693-0819-6.ch005
OnDemand:
(Individual Chapters)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Cloud computing involves virtualization, distributed computers, networking, software, and web services. Clouds have servers, datacenters, and customers. It has fault tolerance, high availability, scalability, flexibility, little user overhead, low ownership cost, on-demand services, etc. These issues demand a robust load balancing mechanism. Balanced load distribution improves resource use and task response time by preventing some nodes from being completely loaded and others idle. Load balancers match processor and network node performance. A load balancing solution that improves throughput and latency for application-based virtual topologies with variable cloud sizes will apply the divisible load scheduling theorem.
Chapter Preview
Top

Goals Of Load Balancing

According to what is said in Nabeel et al. (2012) the objectives of load balancing are as follows:

  • •

    Consequently as to significantly raise the level of performance.

  • •

    To have a backup plan in place in the event that the system fails in any way.

  • •

    In order to preserve the reliability of the system.

  • •

    To make room for any changes in the system in the future.

Types of Load Balancing Algorithms

There are three different categories that load balancing algorithms can fall into, depending on who started the process:

  • •

    Sender Initiated: In the event that the load-balancing algorithm is set up and initialized by the sender

  • •

    Receiver Initiated: In the event when the receiver is the one to commence the load balancing algorithm:

  • •

    Symmetric: It is a collection of actions taken by the sender as well as those taken by the receiver.

  • •

    Depending on the configuration of the system at the moment, load balancing methods can be separated into one of two groups, as shown in (Wan, Zhiguo, et al. (2012))

  • •

    Static: It is not dependent on the way the system is operating at the moment. It is required that you have prior knowledge of the system.

  • •

    Dynamic: The current state of the system is taken into consideration while making decisions regarding load balancing. There is no requirement for prior knowledge. Therefore, it is preferable than a static method.

  • •

    In this article, we will talk about a variety of potential algorithms for dynamic load balancing for clouds of varying sizes.

Complete Chapter List

Search this Book:
Reset