Simplifying Stateful apps on Kubernetes: a git like Workflow

A stateful application in kubernetes consumes resources that include persistent data storage, cpu, memory. Managing the stateful application requires storage management of application metadata and application data. A stateful application like postgres, mongodb, mysql, redis etc has pods (containers), secrets (password for db), configmap (environment variables) as a metadata. It also needs a storage volume to store persistent data.

A kubernetes centric storage architecture is needed to power the lifecycle operations of the stateful apps. It will manage the entire state of the application instead of data storage only. The workflow in git (distributed version control) has operations like cloning a project, taking a branch, make changes, rollback, push branch to remote repo and so on. This presentation will delve into the challenges and look into solutions to make a stateful app lifecycle management as simple as a git workflow.

Finally, we present the distributed storage stack and data management orchestration engine that enables git like workflow for the stateful applications that include snapshot, clone, rollback, push app to cloud repo (aws s3, google cloud storage, NFS mount, Azure block blob), pull an app from cloud repo, create a new app from a cloud repo.