Hybrid algorithm of max-min firefly for parallel job scheduling in cloud computing R. Sundar Rajan 1 Kalasalingam University Tamil Nadu, India sunrsss@yahoo.co.in, V.Vasudevan 2 Kalasalingam University Tamil Nadu, India vasudevan_klu@yahoo.co.in, Abstract— Cloud computing is an emerging computing technolo- gy that uses the remote servers to maintain data and application. A parallel job is run as multiple processes. A High-level job is submitted to the job scheduler and after submission is divided into subordinate jobs that run at the same time. In existing sys- tem the two tier priority based parallel jobs more number of jobs will not to be processed and some jobs are waiting in the queue. So Execution time of the job value is increased and Wastage of memory. The long execution time of the job will be waiting in the queue. This paper we proposes a new scheduling algorithm is hybrid algorithm of firefly Max-Min algorithm for scheduling of jobs on the cloud. The computing capacity of each node will be in two tiers. The two tiers will process from queue. The first vm is foreground and second vm is background. The vm will be allocated for equal partition to run the jobs in foreground and background .The parallel jobs to make the minimization of the makespan and flow time of the completing jobs and more number of jobs to run in the virtual machine to improve the re- sponsiveness of these jobs. Keywords- Cloud computing, scheduling, Firefly algorithm, Max-Min algorithm, Makespan, Minimum completion time, Minimum execution time. INTRODUCTION Cloud computing is a general term for anything that in- volves delivering hosted services over the Internet. These ser- vices are divided into three types: Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS) and Software-as-a-Service (SaaS). The name cloud computing was inspired by the cloud symbol that's often used to represent the Internet [1]. Saas is a Software as a service It will be providing the software, hard- ware and platform. The next service is Infrastructure as a ser- vice it will organize to the hardware. Paas is Platform as a ser- vice it will provided by the platform of the operating system. Cloud computing is an internet-based computing the computing will be defined to achieve a super computer operation is called computing. It can be shared the software and hardware re- sources are provided by user. Two categories of scheduling strategies; first one space sharing and another one is time shar- ing. The time sharing techniques virtualizes the physical ma- chine. The space sharing techniques runs the works on different virtual machine at the same time. A. Scheduling A “Scheduling” is the process of controlling and sending the priorities of the message to the CPU. The processor of the operating system of the program is called scheduler. The scheduler is choosing the processes from the list of complete process. Each process have the own correspond id to run the operating system the ready process will be execute the job. The use of the scheduling method increase utilization of the re- sources. Scheduling is used to allocate the jobs to run in partic- ular time to run in particular machine. The majority of the re- searches in job scheduling adopt a paradigm in which a job in cloud computing system is characterized by its workload, time limit and the corresponding value obtained by its completion earlier than deadline, which are factor considered in devising an effective scheduling algorithm. First-Come-First-Serve (FCFS). In this algorithm, jobs are continuous to order of en- trance [2]. The processor will start the execution automatically if it is free. Or else, the first job will be waiting in the queue some currently running job to terminate and waiting job will be run to schedule. So waste of processing time and processors will be idle time of the waiting state. B. Scheduling Jobs based on parallel In this model, jobs have process in parallel, the scheduler to schedule the jobs; one job will be processed at multiple times of the processors in simultaneously. Backfilling: Quinn Snell et al. [2] Proposed a backfilling scheme. Back- filling method is a space sharing optimization the jobs will be run in FCFS order. It will choose the small jobs to run in the head of the queue and remaining jobs will be waiting state or idle state. It will be avoid the situation of the FCFS order is totally violated and some jobs are never run. There are two types of backfilling algorithms: 1. Conservative Backfilling 2. Aggressive Backfilling Conservative Backfilling: Ahuva et al. [3] proposed a con- servative backfilling approach. In this method, the jobs are scheduled to order of the arrival time it is increase the number of jobs. The jobs will be selected by the arrival time so longest arrival time of the jobs will be waiting state then smaller arrival time of the jobs will be scheduled to run in the backfilling con- cept. EASY Backfilling: Ahuva et al. [4] also proposed an ag- gressive approach; provide an order to only the job At the top of the job queue and only allow job at the top of the queue can be pre-empt previous jobs. The user will not be guaranteed with the reply time of the jobs submitted. International Journal of Computer Science and Information Security (IJCSIS), Vol. 14, No. 9, September 2016 220 https://sites.google.com/site/ijcsis/ ISSN 1947-5500