A quick introduction to Velero

#devops #opensource #kubernetes

Introduction

Formerly known as "Heptio Ark"
It provides tools to Backup and Restore K8s cluster
It consists of the following components:
- Server: Runs on K8s cluster
- CLI: Runs locally
Current release: v1.5.x
GitHub stars: 4.6K+

Checkout here for more details

Architecture

Source: https://velero.io/docs/v1.5/img/backup-process.png

Velero supports the following operations:

On-demand backup
Scheduled backup
Restore

Each operation is a custom resource defined with the custom resource definition and stored in etcd. Velero has controllers that act on these resources.

On-demand backup

Creates a copy of k8s objects in the form of tarball and uploads that tarball to cloud object storage

Note: Backup operation is not atomic.

Scheduled backup

As the name suggests you can schedule the backup at a particular time. You can also schedule recurrent backups. Provide Cron expression to specify a schedule. For e.g. Use the 0 0 * * * Cron expression to take backup once a day.

Restore

Creates k8s objects from the definition stored in cloud object storage.

Note: As a good practice always set the mode of backup storage location to Read-Only during restore to avoid update or deletion of backup.

Velero uses cloud object storage as a source of truth. It synchronizes data between cloud object storage and the K8s cluster.

Features

Filtering k8s objects based on type, namespace & label during backup and restore
TTL (Time to Live) for backups. Backup resources will be deleted after TTL is reached
Different lifecycle hooks for both backup and restore operations
Supports namespace remapping i.e. K8s object from namespace A can be restored to namespace B