Study On Key Technology Of The Survivability Assurance Solution For Multi-process Software

With the arrival of the age of big data, computers bear more and more tasks of dataprocessing and analysis, in our daily lives. Internet companies use computer software tohandle a variety of data collected through the network, for the sake of providing a richerand better Internet service. In various fields of scientific research, a large number ofscientific computing tasks require huge amount of computers to accomplish. In addition,the software needs to run continuously for a long period of time in order to produce thefinal results, which may take a few hours, days or even months. During the process ofrunning, the system may be powered off, or collapsed by external application invasions.It will lead to a great loss of intermediate results that have been calculated. The previousapproach is to run the software from the scratch, which can’t be accepted for highrequirements of real-time tasks. This situation makes the survival of the protection ofcomputer software systems more and more important.This dissertation studies the key protection technologies and theories on thesoftware survivability. From the perspective of software engineering developmentprogram, it also gives a detailed description of development schemes for the protectionof software survivability. An added development specification in the softwaredevelopment process is summarized in order to enhance the software survivability. Thedissertation also focuses on another very effective technology of ensuring the softwaresurvivability. It bases on the mechanism of the process checkpoint. This dissertationanalyzes mechanisms which utilize the process checkpoint to protect the survival of thesingle-process software. These mechanisms consist of the implementation theory andtechnology of the three-part operations: checkpointing the process, identifying the faultprocess and recovering the fault process.After the in-depth study of single-process checkpoint mechanism, the dissertationtakes advantage of the existing single-process checkpoint technology to design aprotection for multi-process software survivability solutions, i.e. Assurance System forthe Survivability of Multi-process Software System (ASSS system). The dissertationdescribes the components and the architecture design of the system in detail. This system aims at setting up multi-process checkpoints, identifying fault process andrestoring failure process, and proposes corresponding implementation algorithmsrespectively.At last, after the theoretical design, the ASSS system is implemented withfunctional and performance tests. Based on the analysis of experimental data, thesurvivability Assurance solution for multi-process software proposed in the dissertationis feasible and valid. The ASSS system has better performance and practicability, alsohas some practical value.