A Comprehensive Survey of Load Balancing Strategies Using Hadoop Queue Scheduling and Virtual Machine Migration

Abstract

The recent growth in the demand for scalable applications from the consumers of the services has motivated the application development community to build and deploy the applications on cloud in the form of services. The deployed applications have significant dependency on the infrastructure available with the application providers. Bounded by the limitations of available resource pools on-premises, many application development companies have migrated the applications to third party cloud environments called data centers. The data center owners or the cloud service providers are entitled to ensure high performance and high availability of the applications and at the same time the desired scalability for the applications. Also, the cloud service providers are also challenging in terms of cost reduction and energy consumption reductions for better manageability of the data center without degrading the performance of the deployed applications. It is to be noted that the performance of the application does not only depend on the responsiveness of the applications rather also must be measured in terms of service level agreements. The violation of the service level agreements or SLA can easily disprove the purpose of application deployments on cloudbased data centers. Thus, the data center owners apply multiple load balancing strategies for maintaining the desired outcomes from the application owners at the minimized cost of data center maintainability. Hence, the demand of the research is to thoroughly study and identify the scopes for improvements in the parallel research outcomes. As the number of applications ranging from small data-centric applications coming with the demand of frequent updates with higher computational capabilities to the big data-centric application as big data analytics applications coming with efficient algorithms for data and computation load managements, the data center owners are forced to think for efficient algorithms for load managements. The algorithms presented by various research attempts have engrossed on application specific demands for load balancing using virtual machine migrations and the solution as the proposed algorithms have become application problem specific. Henceforth, the further demand of the research is a guideline for selecting the appropriate load balancing algorithm via virtual machine migration for characteristics-based specific applications. Hence, this paper presents a comprehensive survey on existing virtual machine migration and selection processes to understand the specific application-oriented capabilities of these strategies with the advantages and bottlenecks. Also, with the understanding of the existing measures for load balancing, it is also important to furnish the further improvement strategies, which can be made possible with a detailed understanding of the parallel research outcomes. Henceforth, this paper also equips the study with guidelines for improvements and for further study. Nonetheless, the study cannot be completed without the mathematical an...

RETRACTED ARTICLE: SLA-aware load balancing using risk management framework in cloud

“…All approaches provide the benefits of shorter response time. Nevertheless, the resource duplication cost must also be taken into account as an extra cost (Niladri Sekhar Dey et al 2019). In general, cloud data center networks are based on huge layered switches with a vast volume of data between thousands of servers has to be transmitted.…”

Providing Consistent State to Distributed Storage System

“…In distributed systems, applications requiring large volumes of data are studied under "data-intensive computing", in which most of the processing time is dedicated to reading/writing or manipulating the data. With the surge in the amount and pace with which the data are generated, the focus of distributed systems such as cloud computing has shifted from compute-intensive to data-intensive domain [13]. The data to be processed in distributed systems can be broadly classified into three categories: structured, semistructured, and unstructured.…”

LSTPD: Least Slack Time-Based Preemptive Deadline Constraint Scheduler for Hadoop Clusters

“…Hadoop utilizes a special and reliable shared storage system, called Hadoop Distributed File System (HDFS), to manage data access in a distributed approach. The HDFS makes ease to access to a data block for a MapReduce job to run the corresponding big data application in parallel [7]- [9]. It can provide a stable and reliable interface to the application, and build a distributed system with high reliability and scalability.…”

“…Buyya et al [17] proposed a scheduling algorithm to reduce the response time of the cloud resource with minimum cost. Several scheduling schemes were proposed to balance the workload based on the amount of available resources [9], [18]- [21]. All these schemes are non-preemptive and do not consider priority based on the deadlines of the submitted jobs.…”