In cloud computing systems, current works do not challenge the database failure rates and recovery techniques. In this chapter, priority-based resource allocation and scheduling technique is proposed by using the metaheuristic optimization approach spotted hyena optimizer (SHO). Initially, the emperor penguins predict the workload of user server and resource requirements. The expected completion time of each server is estimated with this predicted workload. Then the resources activities are classified based on the criteria of the deadline and the asset. Further, the employed servers are classified based on the workload and the estimated completed time. The proposed approach is compared with existing resource utilization techniques in terms of percentage of resource allocation, missed deadlines, and average server workload.
TopIntroduction
Cloud Computing
Cloud computing is different from traditional computing, which relies mainly on personal devices. This enables mobile sharing of computing resources. It offers, at minimum time, versatile and limited access to resources. Hardware or software can be the shared tools. Cloud offers various products such as Service Software (SaaS), Service Platform Service (PaaS), Service Infrastructure (IaaS) (Sheetal & Ravindranath, 2019).
The advantages of Cloud Computing are:
- 1.
Applications can be accessed as utilities throughout the web.
- 2.
No specific software installation needed for accessing cloud applications.
- 3.
The PaaS service provides various deployment tools and runtime environments
- 4.
It provides platform independent access to all clients.
- 5.
Supports load balancing (Bhavani & Guruprasad, 2014).
The advantages of Cloud Computing's are:
- 1.
Applications can be accessed throughout the web as utilities.
- 2.
No installation of specific software required to access cloud applications.
- 3.
PaaS provides different implementation methods and runtime environments
- 4.
Provides separate application access for all customers.
- 5.
Supports balance of load (Bhavani & Guruprasad, 2014).
Problem Identification
The resource usage and energy consumption parameters were mainly considered for the allocation of VM in the multi-agent-based VM allocation approach (Wang et. al, 2016). But the work load forecast or future asset needs are not addressed. In fact, there was no understanding of the deadline for each mission (Dhiman, 2020), (Dhiman & Kaur, 2019), (Dhiman & Kumar, 2019).
In (Student, 2015), when a job with high priority (with a low deadline) arrives, the job with low priority (with a high deadline) has been preempted, allowing the job with high priority to be carried out in its capacity. But it does not control the workload of the PMs, nor does it check the size of the resources requested. In (Xiao et. al, 2012), (Choi & Lim, 2016), (Kumar et. al, 2017), the workload is predicted based on the need for future resources. It then migrates the VM from a hot spot to a resource-based cold spot. But the energy cost for the services used was not included in this method. In fact, there was no understanding of the deadline for each mission (Dhiman & Kaur, 2018), (Singh & Dhiman, 2017), (Kaur & Dhiman, 2019).
Some works on cloud-based task scheduling based on ABC are available (Kruekaew & Kimpan, 2014), (Kimpan & Kruekaew, 2016). In the fitness function, they consider the VM load to select the VMs (Hesabian et. al, 2015) considers making pan time and load balancing (these metrics have not been defined) for fitness function (Sheetal & Ravindranath, 2019), (Garg & Malhotra, 2017) also considers completion time and loading as a fitness function for selecting VMs (Garg et. al,2018), (Garg et. al,2019), (Garg et. al,2019).
But the main advantages of the solution over these works are: