Approaches to Fault Tolerance and Disaster Recovery in DevOps Processes

BACKGROUND The purpose of the study was to analyse approaches to improving reliability and recovery from failures in the processes of software development and operation. MATERIALS AND METHODS The research utilised a qualitative methodology, conducting semi-structured interviews with industry experts to investigate fault tolerance and disaster recovery solutions within DevOps processes. Grounded theory coding was applied to analyse the data, identifying key themes that were mapped to the research questions and comparison matrix. RESULTS The results showed that the microservices’ architecture is highly flexible and scalable, which allows for faster adaptation to changes in the workload and reduces the likelihood of system failures. However, the complexity of managing dependencies between services can lead to new problems. Serverless computing can be effective in reducing infrastructure costs and simplifying scaling, but there is a need for detailed monitoring of the functioning of functions and management of their resources. Containerisation and container orchestration are suitable for providing a high level of portability and application isolation. At the same time, the growing complexity of orchestration systems can be a challenge for the team working on their administration. Cloud computing and load balancing should be used to allocate resources and reduce the risk of stress on individual components, but they require constant monitoring to ensure optimal performance and identify potential problems. In addition, data replication and backup provide reliable protection against data loss, but the increase in data volume places new demands on the speed and efficiency of backup processes. Monitoring and alerting can improve timely detection and response to problems but require regular updates and adaptation to new threats. Automated testing and disaster recovery planning have proven to be key to reducing system recovery times, but the complexity of setting up automated scripts and tests remains an essential issue. CONCLUSION Overall, the results indicate the need for continuous improvement of existing methods to ensure increased resilience and adaptability of systems to new threats when working with software.