Design And Implementation Of Instances Scheduling Optimize And Timing Resume&suspend Based On Openstack

With the development of the information society, the desire of IT infrastructure and service of company and people has changed profoundly. The companies demand the 7×24h service, huge scale data store, elasticity compute and high availability, however, the low cost of IT infrastructure and maintaining for more. The Cloud Computing just satisfy the desire completely. Cloud Computing aggregate the computing and store capacity and maintain them, and make them virtual. The users can use the source of Cloud Computing as the model as they use electric and tap water,and cost them on the basis of their demand. The virtual technology can improve the use ratio of the source. The vantage of the virtual has shown obviously when the scale of deployment increasing.Openstack has rapidly developed by the character of open source and supporting by lots notable corporations. It becomes to be the best choice for the companies, governmental agencies, universities to deploy Cloud Computing. Openstack has already provide the cloud store?network virtual?instances manager ?security strategy etc function and service.However, I had analyzed the code of the Openstack Nova schedule which take charge of scheduling instances, I find the schedule module inspect the available source via the static quota only. The static quota cannot reflect the dynamic real-time message of the instances running. Adding the calculation of the available resource excepting the inactive status instances, once the inactive status instances resuming while new instance has been scheduled to the same host would make the host' running resource shortage and lower the quality. To solve the problem, I fix a monitor daemon for each host for monitoring the dynamic resource message and the history of CPU,RAM,network and its like running summit. After getting these messages, it would to provide the function of getting rid of the unsuitable host for Nova' Filter and Weight process and avoid the bottleneck of the host.In addition I have developed a function of timing instance resume & suspend for uers.Openstack could pre-schedule the instances based on users' work paln with setting the instance running of time bucket by it.The function could reduce the users' operation of Openstack and the time of waiting for the instance resume.It could improve the user experience.