Back up Apache Kafka on Kubernetes

Kubernetes-native backup for any Kafka​

The OSO Kafka Backup Operator extends Kubernetes with Custom Resource Definitions in the kafka.oso.sh/v1alpha1 API group. You declare a KafkaBackup or KafkaRestore object, and the operator runs it as a Kubernetes Job that executes the kafka-backup engine. Because the desired state is a custom resource, backups fit straight into GitOps and kubectl workflows.

It targets any Kafka the pod can reach over the network — self-managed brokers on the same cluster, Amazon MSK, Confluent, or Kafka in another namespace. If your cluster is managed by Strimzi, use the Strimzi Backup Operator instead: it resolves bootstrap servers and credentials from your Strimzi resources. This generic operator is the right choice for everything else.

Install and schedule a backup​

Install the operator with Helm, then apply a KafkaBackup. A schedule field turns it into a recurring backup; stopAtCurrentOffsets: true makes each run a point-in-time snapshot. To restore, apply a KafkaRestore that points at the backup and, optionally, a target time.

helm repo add oso https://osodevops.github.io/helm-charts/
helm repo update

helm install kafka-backup-operator oso/kafka-backup-operator \
--namespace kafka-backup \
--create-namespace

apiVersion: kafka.oso.sh/v1alpha1
kind: KafkaBackup
metadata:
name: hourly-backup
namespace: kafka-backup
spec:
kafkaCluster:
  bootstrapServers:
    - kafka.default.svc:9092
topics:
  - orders
  - payments
storage:
  storageType: s3
  s3:
    bucket: my-kafka-backups
    region: us-west-2
    credentialsSecret:
      name: s3-credentials
schedule: "0 0 * * * * *"   # every hour; cron format includes seconds
stopAtCurrentOffsets: true
compression: zstd

apiVersion: kafka.oso.sh/v1alpha1
kind: KafkaRestore
metadata:
name: pitr-restore
namespace: kafka-backup
spec:
backupRef:
  name: hourly-backup
pitr:
  endTime: "2026-07-06T12:00:00Z"
kafkaCluster:
  bootstrapServers:
    - kafka.default.svc:9092

Without the operator​

Not every team wants a controller. The same kafka-backup engine runs as a plain Kubernetes CronJob for scheduled backups or a one-off Job for a single backup or restore — the Kubernetes deployment guide covers both manifests. The operator adds schedule reconciliation, status reporting, and GitOps-native CRDs on top of that same engine, so you can start with a CronJob and move to the operator later without changing how backups are written.