Font Size: a A A

Research And Implementation Of Load Balancing Strategy Based On Openstack Service Cloud Platform

Posted on:2015-02-09Degree:MasterType:Thesis
Country:ChinaCandidate:Q ZhangFull Text:PDF
GTID:2298330467963133Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the popularity and development of Internet, data is keeping increasing and we have stepped into a phase of data expansion. The advent of cloud computing has greatly erased this pressure. Cloud computing can provide "unlimited" resources, which can be rent to remote users according to their demands. Cloud computing has many advantages, such as massive computing capacity, low cost, on-demand use, etc., but it produces a lot of virtual resources, which are hard to manage and control. The uncertainty of users’choices on virtual machine will easily cause the imbalance of resource nodes. So this thesis proposes a resource scheduling framework with load balancing and dynamic scaling character.Resource scheduling framework consists of four parts:historical data center, load balancer, extended decision-maker and resource allocation manager. This framework can achieve load balancing and dynamic scaling. By using load balancing strategy based on capability matching, load balancer can achieve load balancing using statistical forescasting algorithm to predict load, and assign tasks to proper virtual machine whose capability matches the requirement.This thesis mainly focuses on the design and implementation of load balancing strategy based on capability matching. This strategy includes three parts:load forecasting, load monitoring and task scheduling. Load forecasting section uses historical data to do prediction; Load monitoring section monitors virtual machine’s status using munin and collect software; As for task scheduling, this thesis uses request-capability matching algorithm to dispatch each task. At last, we conduct some experiments and tests on resource scheduling framework, and compare our load balancing strategy with traditional static and dynamic strategies. The results show that resource scheduling framework can achieve good load balancing and dynamic scaling, the load balancing strategy based on capalibity matching can better adapt to dynamic changes of workload and use resources more properly.
Keywords/Search Tags:cloud computing, load balancing, load predictiontask scheduling
PDF Full Text Request
Related items