1.1 Load Balancing
Load balancing (Wang, 2010),(Gupta,2016) is the concept of dispensing the task to the set of computers in a group of systems (network). It would help to enrich the efficiency of the cloud environment by decreasing both the time of execution, Makespan, and increasing the throughput. It also avoids too much overload and reduces the total waiting time of the resources. So, it is defined as the procedure for shifting the workload among the processors by enriching the performance of the system. For achieving maximum client satisfaction, the distribution of the workload for the higher resource utilization ratio is an advantage of load balancing. The primary aspiration is to attain high user satisfaction and high resource utilization ratio, that is done by the load balancer, which dispenses the dynamic workload amongst different systems or nodes in the cloud environment so that a single system or node is not overburdened and hence enhances the overall performance of the node or system. However, the fact is that load balancing must be done in a way that would help to attain optimal resource utilization and reduce resource consumption. There are various qualitative metrics and parameters for balancing the load in cloud computing (Kaushik,2019) that are as follows:
- 1.
Throughput: Throughput is the total number of tasks executed in a given amount of time for better performance of the machine, throughput must be high.
- 2.
Associated Overhead: For executing the load balancing algorithm, the amount of overhead produced should be minimum.
- 3.
Fault-tolerant: The load balancing algorithm should be able to execute well, even in the case of failures.
- 4.
Migration time: Migration time is the duration by the job to migrate from one machine to another, and it must be lowest for improved performance of the machine.
- 5.
Response time: Response time is the time to execute the specific task in a minimum amount of time.
- 6.
Resource Utilization: Utilization of the resources sufficiently improves the performance of the machine. Resource utilization refers to the number of utilized resources.
- 7.
Scalability: Scalability means to scale up and down the resources as would be required by the load balancing algorithm.
- 8.
Performance: The main factor is to increase the performance of the system. The increment is only possible if and only if all the above parameters are satisfied.
Also, there are some challenges (Alam, 2017,), (Aslanzadeh, 2011), (Chaczko, 2011), (Desai, 2017), (Grobauer, 2011) in load balancing that are as follows:
- 1.
Automated Service Provisioning: The main challenge in load balancing (Velagapudi, 2014) is the automation of the resources, which can be allocated or released mechanically according to the need.
- 2.
Virtual Machine Migration: Virtual machine migration is the challenge of load balancing because it means moving the task from the overloaded machine to the under-loaded machine. While migration takes into account certain parameters and there are also different kinds of virtual migrations. The primary goal of a virtual machine is to allocate the task in a data center or set of data centers.
- 3.
Energy Management: Energy saving and energy management (Kansal, 2012) are one of the crucial challenges for load balancing algorithms. To achieve green computing in cloud environment, energy saving is extremely needed. There is a constant need for an energy-efficient algorithm that reduces resource consumption but maintains fine performance.
- 4.
Stored Data Management: There is an exponential growth in the last few years due to a large amount of data that needs to be stored. This data may belong to any corporation or any person. A more significant issue for cloud computing is the management of data storage for a corporation or person. Subsequently, the question arises of distributing data to the storage for better storage of fast access data.
- 5.
The Emergence of Small Data Centers for Cloud Computing: Due to more advantageous, inexpensive, and less energy consumption, small data centers have some profit over large data centers. Leading to geo-diversity computing, small providers can render cloud computing services to make specific and sufficient response time with an optimal sharing of resources, balancing the load would become trouble on a worldwide. So, for improving the overall performance of the machine, load balancing is used as:
- a.
To distribute the higher processing load to lower processing nodes.
- b.
To disseminate the load evenly amid the nodes.
- c.
To attain a high satisfaction of user & proper utilization of resources.