Research And Implementation Of The Checkpoint Technology In High Availability System

With the computer further applied in all fields of human living, many new computer applications and services arise continuously and spread widely, moreover the data increases faster than application itself as exponential, Research organizations and companies focus the high availability computer system. The requirement to design a new kind of high availability system is increase. The availability of the computer system is composing with the availability of data, software and hardware, the scheme of the high availability cluster system based on storage network system is try to increase the availability of data, software and hardware, at the end to realize the high availability of the computer system.The computer system is constituted by data, software and hardware, the characteristic in high availability of these three elements is unlikeliness, but these three elements depend on each other and are with each other for complement. Using the storage network system to guarantee the data's high availability, making use of checkpoint and process migration technology to implement the high availability of software, utilizing error-detection technology, hot-backup technology and cluster technology to improve the high availability of hardware. Synthesize the consideration, putting forward the high availability cluster system scheme, which is based on network storage system.Error-detection and error-diagnosis is the fundamental technology of the system, which would to recovery from the fault state. Error-diagnosis should position the fault accurately and on time, at the same time should have a small system expense. The detection module has an open framework scheme, which can make use of every kind of agent. Detection agents make use of checkpoint technology and process migration technology to detect the process state, recovery the process from the fault state.Inside the implementation and improvement of the checkpoint, the scheme supports multi-processes and is based on service group. The implementation of the system is in a shared storage network environment, this solution enhances the I/O performance and the reliability of the checkpoint file, makes the system more scalable. In order to increase transparency to the process and decrease run-time overhead to the system, this solution adopts a dynamic adaptive mechanism which adjusting the checkpoint's time interval.