LOAD BALANCING TECHNIQUES IN CLOUD COMPUTING

: In today’s era, distributed computing is rising in the field of data processing. It gives a broad measure for the services to be processed for clients through the web. Everyone’s data is on cloud. There resides many problems when we process data on cloud like as security, privacy etc. And load balancing is one of the problem in cloud computing. There are many requests generating on a single virtual machine and these requests are to be processed on various machines. This is controlled by load balancer. Load balancing is the technique used for assigning different-different tasks on different-different virtual machines by using various algorithms. Load balancing plays a vital role in the era of cloud computing which is assumed to circulate on various machines. This paper presents the study of various load balancing techniques.


INTRODUCTION
Cloud computing is a paradigm used for storage capacity, computation and services among massive users. In essence, cloud computing [1] overlaps many existing concepts such as grid, distributed and utility computing. Big cooperates like Google, Amazon and IBM offering services evolved these concepts and are focusing on Cloud Computing. If cloud computing steps into our daily life then our information such as our email, spreadsheet, word processing which we locally store on our desktop can be stored remotely on various virtual machines residing on cloud. Then we can access these services at any time from any terminal. Due to these characteristics cloud computing is becoming more attractive to people. The cloud in cloud computing can be defined as the set of hardware, networks, storage, services and interfaces that combine to deliver aspects of computing as services [2].  [11] proposed a hybrid approach of encryption techniques. The main benefit of the hybrid scheme is to give more security in the cloud. The hybrid algorithm is based on the password and security key. In this case the client is free from the fear that the administrator knows the password. The password and security key are saved in database. [12] proposed improved load balanced min-min (ILBMM) algorithm using genetic algorithm (GA). It minimizes the make span and increases the utilization of resources effectively. [13] shows a comparison to evaluate various existing load balancing techniques. By this comparison, performance can be improved. The various load balancing techniques are analyzed here. [14] presents cloud computing architecture, virtualization, load balancing issues and various load balancing algorithms. Load balancing is a technique which provides methods to maximize output, use of resources in efficient manner and performance of the system. [15] presents a meta-heuristic approach of ACO algorithm to solve the task scheduling problem in cloud environment. Comparative analysis shows that proposed LB-ACO provides better results.

III. Comparative analysis
Year The power consumption cost was reduced.
Doesn't work for large data centers. 2011 Join-Idle-Queue.
Balance load across dispatchers and then assign jobs to processors.
Reduce system load. More power consumption.

2011
Randomized. It randomly assigns jobs to VM. It is very simple algorithm.
Current load is not considered. 2012 AMLB algorithm.
Request is allocated to least loaded VM.
Consider availability and also load of VM.
Processing power of VM is not considered. 2012 Honey bee foraging.
Decentralized honey based load balancing technique.
Balances load efficiently.

Migration communication delay was not considered. 2013
LBIMM If calculated completion time is less than make span of min-min then it is reassigned to resource.
Reduces the overall completion time.
Doesn't consider the priority of jobs.
Similar to Min-Min except that large jobs will be executed first.
Reduces the make span. Smaller jobs have to wait for a long time. 2015 Memetic algorithm Local search algorithm which are invoked according to search process.
Local search method is good for solving these problems.
Such algorithms are CPU insensitive.
Weight is assigned to each VM based on its processing capacity.
Better resource utilization.
Processing time of each request is not considered.

IV. CONCLUSION
Load balancing is a big challenge in cloud computing. It is difficult to maintain to make idle services or to fulfill all the required demands. So there is distributed environment always in need. The components are present throughout the wide area network. Jobs cannot be assigned to the appropriate servers and clients individually efficiently. Load balancing algorithms are classified into two categories as static and dynamic algorithms. Static algorithms are suitable for homogeneous and stable environments. They produce very good results in these environments. But they are not flexible and do not match up with dynamic algorithms in execution prospective. Dynamic algorithms are more flexible and are better prior to as well as during the run time of tasks. Load balancing is the process of improving the performance of system through the redistribution of load on various VMs. In future work, ant colony optimization will utilize to tackle all the problems in cloud computing. ACO works distributed and in parallel. Hence they distribute load in an efficient manner on cloud.