This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The Work load management is to produce maximum resource usage, minimize the response time and prevent overload between the system.The Workload include both Transactional and Long running calculation jobs,which is a similar Combination attend of new performance management in different nature of a heterogeneous set of mixed work load on the same physical hardware.The Resource utilization is a challenging goal in presence of heterogeneous workloads in cloud computing Environment.We propose Well known Optimization algorithm Harmonic Search Optimization algorithm.Our algorithm to find the optimal solution based on the placement matrix when calculate the mixed workloads and generate best new placement vector that enhances accuracy and convergence rate of harmonic memory.
Keywords - Workflow,PerformanceManagement,Workload management,Placement Vector
Cloud Computing support on demand self service,pay-for-use and dynamically scable storage sevices over the internet .Cloud computing offer three fundamental services Infrastructre as a Service(IaaS),Platform as a Service(PaaS),and Software as a Service(SaaS).In the IaaS Provides Infrastructure Components like Storage,firewalls,Networks, Loadbalancing and other computing resources to their Customers.IaaS also referred to Hardware as a Service(HaaS). Load banancing is a process of applying the total load to the individual nodes of the shared system to enhance the response time of the job and to make effictive resource utilization,at the same time extract a state in which some of the nodes are overloaded while some others are under loaded.System load view in terms of CPU load,Delay or Network load and amount of memory used.Some of the goals of load balancing to improve the performance considerably,to have a backup plan when system fails and to maintain the system durability.
Many Organizations to handle Transcational applications and batch jobs are used to deliver services to costomers .Integrated management of these various workloads on same physical server is major challenges on the cloud Environment and need scheduling mechanisam for improving resource utilization to particular workloads.Performance objective for Changed workloads be likely to be of different types.For interactive workloads,objective are basically defined in terms of average or response time with in short time duration,while objective for noninteractive workloads fear the completion time of individual jobs.Different types of workload need changed contorl mechnisms for management.Long running workload require sechduling and resource control management and Transcational workloads are managed using flow control,load balancing and application placement.
This section presents the related work and existing methods used in heterogeneous work load management in Cloud computing Environment.
2.1 Dynamic Application Placement Technique
The dynamic application placement technique has been proposed that allows the management to high level objective of mixed long running and transctional workloads in virtualized environments.To provide performance predictions with value to job completion time on a control cycle which with in the job execution time.At the same time the actual utility complete by a job can only considered at completion time,the technique to predict the utility that each job in the system will finish given pariticular allocation,the concept of hypothetical utility for which they proposed assume that all jobs can be placed all togather,and in which the available CPU power may be allocated among all jobs so that wanted utility is balance among all nodes.
2.2 PALB Algorithm
Power Aware Load Balancing (PALB) algorithm proposed in the previous paper,maintains the state of all compute nodes based on utilization percentage.The algorithm has three basic case.It does collect the utilization percentage of each active compute node,In the case that all compute nodes n are above 80% utilization,all of the applicable compute nodes are in operation or else,the new virtual machine is compute node with maximum utilization.The threshold of 80% utilization was taken since when 20% of the resouces are availbale.The algorithm make decision the no of compute nodes that should be operating.
2.3 Equally Spread Active Execution Load Algorithm
EASE LOAD SCHEDULING
Figure 1. Equally Spread Active Execution load to the cloud Environment.
Here the jobs are submitted by the users to the system,jobs are queued in the stack.the cloud manager calculate the job size and checks the capacity of virtual machine.once the job size and virtual machine size equal,the job schduler allocated resource.The jobs are equally distributed all virtual machine.The advantage of the algorithm to reduce in the virtual machine cost as well as data transfer cost.
2.4 Heterogeneous Workload Management
The Existing system aims to make equal placements in terms of Relative Performance function.The use of RPF provide uniform workload particular performance models that allow proper placement decisions across various workloads .Optimization Objective of given request from various customer applications.Here Relative Performance Function value calculated every application.The main objective of Application Placement Controller is to find the finest feasible placement applications as well as equalient load distribution that optimizes the global objective of the system.Application placement Technique is used to manage Long Running jobs.Limitiation of Existing Systems are the goals are basically defined in terms of average response time or throughput over a short time interval,Non interactive workloads typically require calculation of a schedule for an extended period of time and the cost of the infrastructure and usage of electrical energy is high.
Harmonic Search Optimization Algorithm
General Harmonic Search Optimization Algorithm
Step 1: Initialize random harmonic vetors that will represent the sloution for mixed workloads in cloud environment
Step 2:Devise a new harmonic vector through Random Selection,Memory requirment and pitch Adjustment.
Step 3: Esimate the harmonic using the fitness function for each cell Pm,n.
Placement Matrix(P) of applicaions on nodes cell Pm,n Represents the number of instances of application m on a node n. Evaluate the Harmonic memory and compare it to the New harmonic memory.
Step 4:Step 2 and 3 are repeated until a series of fintie number of invention is achieved.The smallest value or the best harmonic in the harmonic memory is selected to become the solution.
Step 1:Initialize the Harmonic memory parameters Harmonic memory accepting rate(Hmraccept) ,Pitch adjusting rate(Par).
Step 2: Generate harmonic memory with random placement vector accroding to uniform distribution of each workloads
Step 3. Workload Scheduling parameter initialize
Memory ,cpu speed,lengh of work schedule
Step 4: Generate a new placement vector For each workloads
With probability hmcr(harmony memory considering rate)0≤hmcr≤1, Pick the stored value from HM
With probablity 1-hmcr,pick a random value with in the allowed range.
Step 5: Perform additional calculation needed if the value in Step 4 came from HM
With Probability Par(0≤Par≤1)
With Probability 1-Par,do nothing.
Step 6:If new placement vector is better than worst vector in HM,Relace Worst with new placement vector.
Step 7: Repect from Step 2 to Step 6 until termination criterion (e.g. maximum iterations) is satisfied.