Design And Realize Of Cluster Based On VSR Platform

With the maturity of the public cloud services, Now more and more enterprises begin to apply the most of business software and data center to the public cloud, it requires the enterprises to get the resources based on their own needs, and planning their own virtual private cloud on the platform. VSR is a virtual router device. It can make public and private cloud interconnection, building security unified enterprise hybrid cloud. But the typical public cloud platform multi-tenant environment faced by many present situations, enterprises face the management complexity and low reliability on cloud service platform, this thesis studies the design of cluster on the platform of VSR.Through the analysis of public cloud situation. expounds the emerges background of clusters, this paper introduces the basic concept of the cluster, working principle and related technology. Designed a virtual cluster scheme based on VSR platform. The core idea of virtual cluster technology is get multiple virtual router in network layer to be a distributed device on a logical view. This scheme contains cluster establishment and maintenance. By interaction of control packet through control channel during the cluster establishment process, Get a device as the main equipment in a same field between many devices by protocol. So the main device can responsible the operation of the cluster, The management and maintenance of system, the others devices deal with the business as the standby device. eventually create a list of topology in every device.the topological list contains details of all the devices in the stack. after stack formed at the same time, the user can log in system through any devices in the cluster, unified management of all the Members of the equipment, for other devices in the system and network management, the whole stack system is a network node, simplifies the network topology, reduces the complexity of management. The maintenance of cluster is to maintain two different situations. One situation is to maintain the new device can join in the cluster, The other situation is to maintain the master or standby device when the device leave from cluster, stack can make each device in the list of topology information to synchronize updates, thus increasing the robustness of the cluster, Finally analysis the cluster module state machine in the system,Also introduces the process of data when dealing with the message in various states.Passed all the test cases through a series of test methods. Including the formation of the cluster, It includes a single device and two devices, the leave of device in the cluster, including the situation of reboot and power off, new device Join the cluster and form a new cluster. Although meet many problems during the process of the test, all the problem has been solved successfully, the cluster has been delivered finally.