Skip to main content Skip to navigation Skip to footer
MULTI-CLOUD

Fork or move a running Kubernetes cluster

Your cluster moves as one set: nodes, config, and data. To another machine or another datacenter. The cutover is measured in seconds, not a weekend.

Velero restores backups. Cluster API clones shapes. Rediacc moves the running cluster, data included.

Start free trial See the measured numbers → →
rdc cluster fork prod --tag staging
✓ Cloning node images (copy-on-write)........... done ..
✓ Cloning Ceph RBD volumes...................... done ..
✓ Rewriting node identity for new IPs........... done ..
✓ ✓ Fork complete in 46s: prod-staging is up. Parent untouched.
rdc cluster migrate prod --to fsn1-dc2
✓ Pre-copy while the cluster keeps running...... done ..
✓ Cutover: stop, final delta, restart........... 16s ..

Illustrative output; actual runs may include extra logs. CLI reference: rdc cluster migrate

46s
Whole 2-node cluster fork
16s
Migrate cutover, measured
1-5s
Namespace fork
THE PROBLEM

Your cluster is stuck where you built it

Kubernetes made your apps portable. The cluster itself is not. The data lives in one datacenter. Moving it means a rebuild project: new nodes, new storage, restore from backup, retest everything. So most teams never move. And their provider knows it.

27% of cloud spend gets wasted every year Flexera 2024 State of the Cloud Report [1]
17% average amount companies blow past their cloud budget Flexera 2025 State of the Cloud Report [2]
$0.09/GB in fees to move data out, built to make leaving costly AWS/Azure/GCP Pricing 2024 [3]
THE OLD WAY
Month 1 Stand up the new cluster
Month 2 Rebuild manifests and secrets
Month 3 Restore volumes from backup
Month 4 Retest every app
Month 5+ Cut over and hope
WITH REDIACC
Pre-copy
16s cutover
Live
HOW IT WORKS

The whole cluster is one set of files

1

Cluster images plus volumes

Each node runs from an image file. Each volume is a Ceph RBD image. Together they are the whole cluster: apps, config, and data.

2

Fork in place

One command clones every image, copy-on-write. A 2-node cluster forked in 46 seconds in our lab. The parent keeps running.

3

Migrate with a short cutover

Rediacc pre-copies while the cluster runs. Then it stops, sends the final delta, and restarts on the new machine. Measured cutover: 16 seconds.

prod (Datacenter A) Running
Nodes 2 (k3s + Ceph)
Namespaces 8 apps
Volumes Ceph RBD
State Keeps running
Parent Untouched
Fork
46s
prod-staging (Datacenter B) Running
Nodes 2 (k3s + Ceph)
Namespaces 8 apps
Volumes Ceph RBD clones
State Own identity, new IPs
Fork 46s
UNDER THE HOOD

Why a cluster can move like a file

Every node runs from an image file on btrfs. Every volume is a Ceph RBD image. Copying a cluster means cloning those images, copy-on-write. No export, no conversion, no restore step. That is why a fork takes seconds, not months.

Backup tools restore into an empty cluster, piece by piece
Node images and volumes move as one set
Data needs a separate replication product
Ceph RBD clones ride along, copy-on-write
A cluster clone means re-provisioning and redeploying
A 2-node cluster fork took 46 seconds in our lab. The parent keeps running
Cutover plans span a weekend
Measured cutover: 16 seconds. Workloads restart on the destination
WHY IT MATTERS

Keep your Kubernetes. Gain mobility.

Keep your Kubernetes

Rediacc runs k3s for you, or connects to a cluster you already have. Volumes ride on ceph-csi RBD. Your manifests stay yours.

Real datacenter mobility

Provision on your own KVM hosts or on clouds like Linode with private VLANs. Node pools keep disk-heavy Ceph nodes apart from cpu-heavy Kubernetes nodes.

One mentality for everything

A Kubernetes app is a namespace. Fork it like any repo: 1 to 5 seconds, data included. Same commands you already use for Docker repos.

THE GAP

Nobody else moves a running cluster with its data

Backup tools stop and restore. Cluster API clones the shape without the data. RBD mirroring copies volumes, not the cluster. Rediacc moves all of it.

Capability VeleroKasten K10Cluster APIRBD Mirroring Rediacc
Backs up apps and volumes ✓[4] ✓[5] ✗ Volumes only[9] ✓
Moves a whole cluster with its data Stop and restore[6] Stop and restore[7] Shape only[8] ✗ ✓
Forks a running cluster, data included ✗ ✗ ✗ ✗ ✓
Clones one app (namespace) in seconds ✗ ✗ ✗ ✗ ✓
Copies data without a full rewrite ✗ ✗ ✗ Async copy[10] ✓
Short measured cutover for a planned move ✗ ✗ ✗ ✗ ✓

Give your cluster an exit route

Fork it, test the move, then migrate for real. Start your 14-day free trial.

$ rdc cluster fork prod --tag staging

Short on time?

Skip the deep-dive. Grab the five-minute version your team can read at a stand-up.

Download short brief (PDF)
Runs the apps you already ship
Databases, CI/CD, CMS, monitoring, logins, storage. If it runs on Kubernetes, Rediacc can fork it and move it, data included.

Explore Other Solutions

multi cloud

Kubernetes Cluster Mobility

Fork or move a running cluster, data included

Current page multi cloud

Vendor Lock-in Freedom

Your infrastructure. Your choice. Always.

→ multi cloud

Cloud Outage Protection

When AWS goes down, you don't

→ multi cloud

Failover Testing

Test your failover before you need it

→ encryption

Migration Safety

Migrate without risking your data

→ dev env

Environment Cloning

Clone production in 60 seconds

→ backups

Instant Recovery

Full environment recovery in minutes

→ encryption

Encryption You Control

Your keys. Your encryption. No exceptions.

→
Sources & References
  1. Flexera, "2024 State of the Cloud Report," 2024. "27% of cloud spend continues to be wasted." www.flexera.com
  2. Flexera, "2025 State of the Cloud Report," 2025. "With cloud budgets already exceeding limits by 17%, organizations are increasingly turning to managed service providers." www.flexera.com
  3. AWS, "Amazon S3 Pricing — Data Transfer," 2024. "Data Transfer OUT from Amazon S3 to Internet: $0.09 per GB for the first 10 TB/month." aws.amazon.com
  4. Velero backs up Kubernetes cluster resources and persistent volumes and restores them into a cluster. velero.io
  5. Kasten K10 provides backup and restore of Kubernetes applications, including application data. docs.kasten.io
  6. Velero cluster migration restores a backup into a new cluster; workloads are recreated from the backup point, not moved live. velero.io
  7. Kasten K10 migrates applications between clusters by exporting a restore point and importing it into the target cluster. docs.kasten.io
  8. Cluster API's clusterctl move relocates Cluster API objects (the cluster's definition) to another management cluster; it does not move workload data. cluster-api.sigs.k8s.io
  9. Ceph RBD mirroring replicates block device images between Ceph clusters; it covers volumes, not Kubernetes cluster state. docs.ceph.com
  10. Ceph RBD mirroring is asynchronous, based on journaling or periodic snapshots, and is designed for disaster recovery. docs.ceph.com
Fork and migrate timings were measured end to end in Rediacc's KVM test lab on a 2-node cluster. The default consistency is crash-consistent, the same semantics as a power cycle. Workloads restart on the destination during a migrate cutover.