Research On SLA-based Optimizing Resource Management Mechanism In Cloud

Since the concept of cloud computing was put forward, this emerging technology is becoming the research hotspot of industry and academia field. It is also one of the core technology to the application of big data, cloud computing support the large data processing as the bottom. It is a new computing model based on the Internet, sharing resources through virtualization approach, scheduling storage, software, networks, computing and other resources by internet, achieve ondemand consumption and expand. In this service mode, all resources are virtualized as a resource pool, cloud service providers build large-scale data centers to manage and allocate resources to users.Facing complex dynamic resource requirements, how to build an efficient resource management mechanism, has become research focus of academic. On the basis of existing literature, Against the increasing cost of unreasonable resource allocation, low resource utilization and load imbalance. This article, with the goal of maximizing profits, designed a service level agreement(SLA) between the user and the supplier, resolve discrepancies between resource requirements and resource availability, reduce the rate of protocol violation and avoid the hostnode failure, ensure Qo S requirements of users, primarily through the SLA properties: computing power, network bandwidth(transmission capacity) and storage space to measure.Therefore, we propose an efficient resource management mechanism based on SLA, analyze and optimize the resources allocation and virtual machine migration issues in the system.Therefore, the paper’s main research works are as follows:(1)Against the low efficient resource allocation issue in data center, construct a mapping model of user requests and resource allocation, the three properties of SLA as passed parameter: Mips, BW, Memory, call improved intelligent algorithm to allocate resource.(2)In above-mentioned process of resource mapping, with the goal of maximizing profits, design appropriate utility function, to avoid falling into local optimal solution, improved standard PSO algorithm, proposed hybrid particle swarm SA-PSO algorithm to optimized allocation strategy.(3)In cloud environment, faced with a large number of dynamic computing needs, after the resource scheduling, due to load imbalance causes the virtual machine node failure, a selfcontained virtual machine live migration strategy based on colony algorithm was proposed. Reference the ant intelligent searching ability, deduce formulas of resource load of each node. The load value is evaluation function of ant colony algorithm to select the target node migration. Achieve automatic virtual machine migration strategy, eventually form a SLA-based efficient cloud resource management mechanisms.