Extension of a Cloud-Native Kubernetes Backup System for Cross-Region Restore With Disk Snapshots

Abstract

Organizations have increasingly adopted Kubernetes to help manage their applications in a containerized way. On a large scale, these environments become exponentially complex, with the need to hold persistent data. Therefore, reliable backup and restore methods, which can also back up the used volumes' data, are critical. Various tools have been developed to help tackle this need, but they often do not seamlessly integrate with the dedicated storage systems used. This diploma thesis attempts to bridge the gap between one of the most widely used backup and restore tools, Velero, and a popular block storage system for Kubernetes, Longhorn. While some methods already help us integrate Velero with a cluster that utilizes Longhorn, there is no dedicated plugin to integrate these. By developing a dedicated Velero plugin for Longhorn, we help automate the backup of Longhorn disk snapshots and extend its capability to also transfer this data to an external S3 bucket, providing a scalable, cloud-native disaster recovery solution. This work aims to bridge the gap between Kubernetes storage management and cloud-based backup solutions. It thus contributes to the development of more robust and automated data protection mechanisms in cloud-native environments, enhancing both operational efficiency and data resilience for Kubernetes users.