International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 06 | June 2022
p-ISSN: 2395-0072
www.irjet.net
A SURVEY ON STATIC AND DYNAMIC LOAD BALANCING ALGORITHMS FOR DISTRIBUTED MULTICORE SYSTEM Dr.G.Muneeswari1 1Professor, School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In today’s world with large number of super
performance metrics associated with the load balancing algorithms. Finally section 4 concludes the paper.
computers, it’s essential to distribute the load efficiently among different cores among the multicore processing with proper load balancing algorithms. The core idea is that none of the multicore elements should be kept in the idle state. Inorder to implement the efficient algorithm, the better understanding of the distributed system must be dynamically known which actually a challenging task is. Modern framework for load balancing methodologies are integrated with the application itself which can balance the load based on the application level knowledge. But the real scenario lies in the work of processing elements capabilities. So, these algorithms may lead to incorrect simulations. This paper focusses on the comprehensive study of existing load balancing algorithms implemented on the various distributed environments.
2. LITERATURE REVIEW
In this method, the performance (Motwani & Raghavan [1], Zhong Xu & Rong Huang [2] of all the processing elements are calculated before starting any task. This distributed multicore system follows the strategy of master slave processing. The master node allocates the task to various nodes based on the estimated load and arrival time of the tasks to the ready queue. Sometimes the slave nodes will be uded to calculate the work load of the incoming requests. In the McEntire et al [3] algorithm, a designated computational element selects a site for the execution of the new task. Whenever the task is created, a lightly loaded processing element is chosen depending on the complete picture of workload. Depending on the tasteful information, the load balancing algorithm quickly decides upon the load allocation and migration.
Key Words: Load balancing, static, dynamic, resource utilization, Multicore, distributed system
1. INTRODUCTION The distributed multicore system can be homogeneous or heterogeneous with different architectures. A process allocated to the multicore processing elements can be preemptive or non-pre-emptive. A proper and efficient load balancing algorithm must be incorporated mostly in the case of homogeneous system for better resource utilization. In the traditional system mostly sender initiated and the receiver initiated algorithms are implemented which may not be useful for the current massive parallel technology. A new novel hybrid algorithms which can combine the feature of static and dynamic allocation is required for keeping all the processing elements in the busy state.
Unlike the static kind of algorithms, the dynamic algorithms proposed by Malik[4], Wang & Morris[5] allocates the tasks dynamically when one of the processors becomes underloaded. In [6], the routing for hybrid wireless network becomes complicated with respect high transmission rate. When the intermediate hops are preoccupied with data then the load balancing algorithm find out a different route based on the estimated data load during data transfer. The load balancing [7] for cloud will be extremely complex as there are multiple nodes requested at the same time for client data access. A novel methodology related to the workload distribution is implemented on the RAID systems for the better throughput and performance.
The load can be divided based on the implementation of the algorithm. Ideally, underload, overload thresholds are fixed at the centralized master node depends on the expected number of process arrival and capability of computational elements speed of execution. In this paper, a two variant of static and dynamic algorithms are reviewed for various cloud based multicore environments.
IOT systems[8] have less memory and less power active devices wherein the real challenge is to find out the load of data transfer in web based IOT application. Data transfer to the web servers are implemented with the help of traditional sender oriented load balancing algorithms. Mohammed et al.,[9] proposed a server consolidation methodology for
The organization of the paper is given as follows: Section 2 describes the literature review and section 3 explains the
© 2022, IRJET
|
Impact Factor value: 7.529
|
ISO 9001:2008 Certified Journal
|
Page 2535