Dynamic And Reusable Scheduler For Cloud Infrastructure

With the rapid development of cloud computing technology, many IT enterpriseshave launched their own cloud platform. Cloud infrastructure assigns resources made upof CPU and memory to the virtual machines and deploys the virtual machines on thespecial physical machine. The scheduling module used by the cloud platform is built in thesystem, and the scheduling module is poor of flexibility. The developers could not beapplied to customize a suitable policy for their own services according the features of theapplications. Furthermore, the interfaces of the scheduling modules are fixed and ofdifferent in different cloud platforms which lead to a pool of graft.In response to these issues and in light of that the traditional cluster schedulers usemulti-queue, multi-user approach to manage jobs, the dynamic and reusable schedulingsystem for cloud infrastructure service introduces the advantages of cluster schedulers tothe virtual machine scheduling in cloud. The use of plug-in technology makes thestrategies easily to be managed and maintained. The new custom strategy could beimplemented conveniently. And the new strategy can be inserted into the system flexibly.In addition, the system provides a group of unified interfaces to the cloud platform, whichmake the system can easily access different cloud platforms and be reused in differentcloud platforms. The new custom policy could be used in different cloud platforms.Based on Linux operating system, the dynamic and reusable scheduling system forcloud infrastructure service is implemented adopting programming language Python andShell. Testing result shows that the scheduling polices of the system could be dynamicallyconfigured and could be reused in different cloud platforms such as Crane, OpenNebulaand Eucalyptus. When the number of the VMs per minute respectively is1,4,8,10,15and20, the growth rate of the resource utilization is corresponding to0.30%,4.28%,8.74%,10.96%,14.52%and17.01%. The dynamic and reusable scheduling systemenhances the resource utilization well.