Article Preview
TopIntroduction
Cloud computing is an internet-based computing service. Cloud service providers provide the computing resources, but sometimes over demand of resources can cause load imbalance. Proper distribution of workload among the VMs helps to maximize performance and reduce the cost of resources. Researchers use the Honey Bee, Genetic, Adaptive, Agent-based, Ant-colony optimization and other numerous mechanisms to manage the load imbalance. An effective algorithm uses the a) Transfer policy b) Location policy c) Information policy, and d) Selection policy to manage the load balancing in cloud computing (Jennings & Stadler, 2015):
- •
Transfer Policy: Based on threshold state of the CPU. Threshold state shows the high and low load state of VMs. High threshold indicates that no more jobs are executed by CPU and jobs should be migrated to low load VM. Low threshold state shows that more load can be executed by current CPU and ready to take a load from high loaded VM;
- •
Location Policy: Store the information of User Base (UB) and VM locations, regions which are hosted in Data Centers (DC). Location policy helps to find the high loaded and low loaded VM for transferring the jobs. An efficacious location policy minimizes the migration time of jobs from high load to low loaded VM;
- •
Information Policy: Collect the information about available resources in VMs and categorize them on high and low loaded VMs. Information policy periodically update to find the current load status of available VM. Effective information policy helps to manage the load imbalance;
- •
Selection Policy: Is responsible for transferring the job from high loaded VM to low loaded VM. Selection policy reduces the Service Level Agreement (SLA) violations and maximizes the Quality of Service (QoS) (Toosi, Calheiros, & Buyya, 2014).
We map the efficiency of proposed Dynamic Weighted Live Migration (DWLM) algorithm from Push-Pull ((Forsman, Glad, Lundberg, & Ilie, 2015) and Equally Spread Current Execution Algorithm (ESCEL) algorithms ((Bhagwaiya, & Raghuvanshi, 2014). The mapping metrics are migration time, throughput, scalability, availability and reliability:
- •
Migration Time: It is the amount of job migration time from high to low loaded VM. Minimum migration time increases the QoS of cloud providers. Migration time of jobs from host to host must be lowest;
- •
Throughput: Number of tasks executed in specific amount of time. High throughput maximizes the performance of the system;
- •
Scalability: Ability of the algorithm to map the computing capacity of finite VM and also map the utilized and unutilized resources for better load management (Kansal & Chana, 2012).
Recovery from the failure is known as fault tolerance which is mapped by availability of the resources and reliability of the system:
- •
Availability: It is the ratio between the uptime and sum of the uptime and downtime of the system;
- •
Reliability: It is the consistent performance of the system without any interruption. Reliability increases the QoS and minimizes the SLA violations.
Compute, Network, and Storage are the resources of the Cloud computing. A cloud service provider manages these resources by load management mechanism. Researchers are working simultaneously to manage load imbalance. Our proposed work is focused on solving the computing resource load imbalance. Experiment result shows that the proposed algorithm effectively manages the computing load imbalance and gives the best result in Cloud metrics (Jennings & Stadler, 2015). The proposed DWLM mechanism uses the transfer, location, information, and selection policies to manage the load imbalancing, and to map the efficiency of proposed work, we used the cloud metrics.