Back up and recover
your AI pipelines
from a single control plane.
Plakar is the
open-source standard
for efficient, secure backups.
Plakar Control Plane adds
the visibility your organization needs across every inventory. It orchestrates
your entire resilience posture, with backups defined by hand or as code, from one place.
Backup GitLab
App
GitLab
Policies: Production, Operation · app-server-01
Destination store
Restore GitLab
Recovery point
Restore to
Because breach is no longer a risk, it's a certainty.
State-led ransomware, AI-driven exfiltration, and the erosion of the internal perimeter have turned cyber-resilience from a nice-to-have into the most important line of defense.
So "recoverable" has to actually mean something.
Five things Plakar delivers for security and infrastructure teams.
Radically efficient, by design.
Plakar deduplicates and compresses everything before it leaves your system, collapsing the storage and egress overhead that makes legacy backup so expensive. Protecting the whole estate costs a fraction of what traditional tools charge, so you never have to choose what to leave unprotected.
Zero-trust, and tamperproof.
Your backups are encrypted before they ever leave your system, so not even the storage provider can read them. On object-lock storage they're immutable: ransomware and rogue operators can't alter or delete them. Whoever reaches your backups finds only ciphertext they can't touch.
Your data, never held hostage.
Autonomy is more than open source: it's the guarantee that a usable copy of your data always stays under your control, ready to restore or move without asking anyone's permission. Open under a permissive license, self-hostable everywhere, and stored in a non-proprietary format, Plakar turns that control into real leverage. Switch SaaS or cloud providers on your terms, and choose exactly where your data resides to meet local regulation. No appliance lock-in, no vendor-mediated recovery path.
Cloud-native resilience, as code.
Plakar's operator turns Kubernetes into the control plane for the resilience layer of your entire estate, not just your clusters. Define backup, retention, replication, and recovery policies as Kubernetes CRDs, Terraform resources, or Ansible playbooks. Version them in Git, reconcile state continuously.
apiVersion: task.plakar.io/v1alpha1 kind: ScheduleBackup spec: periodicity: 900 # every 15 min source: name: postgres-prod store: name: s3-immutableVersioned in Git · reconciled by the operator
Programmatic by design, for humans and AI.
As more code is written and operated by AI, how easily a tool can be driven matters as much as what it does. Plakar is programmatic by design: every action is exposed through a clean SDK, CLI, and API, so an agent can snapshot the data it works with and restore it the instant something breaks. Docs and source are built to be read, so agents and people drive it right.
It runs on your environment, your storage.
Deploy Plakar on infrastructure you own. One engine protects every source across the estate, written to storage you control.
Run it on your environment
One binary, no appliance, no SaaS dependency. Run Plakar on the infrastructure you already operate: on-prem, in a private or sovereign cloud, or on any hypervisor. Your data and keys stay on infrastructure you control.
Storage provider agnostic
Snapshots land on storage you own, always with zero plaintext: any S3-compatible backend, local disk, or filesystem. On S3-compatible backends, object-lock immutability and append-only keep them tamper-proof.
And your whole backup posture lives as code.
Resilience as Code means your backup posture is defined, versioned, and automated, not clicked through a console. With the Plakar Kubernetes Operator, teams declare their entire backup strategy as native CRDs: sources, storage, schedules, retention, replication, and verification live in Git, reconcile continuously, and ship the same way the rest of your infrastructure does. Manage every policy through the Plakar API, a Terraform provider, or Ansible playbooks.
It fits the estate you already have.
Dozens of sources and tools, one engine, and the long tail keeps growing. Missing something? Ask for it.
S3-compatible storage
Object-lock immutable, append-only destinations on storage you own. LiveCloud providers
Agentless VM-level snapshots across your cloud estate. LivePostgreSQL
Consistent full-database snapshots & dumps Live
MySQL / MariaDB
Encrypted, deduplicated database snapshots LiveKubernetes
Cluster-native snapshots, GitOps-ready LiveProxmox
Agentless VM & container snapshots LiveVMware vSphere
Agentless VM protection for hybrid estates LiveDocker
Snapshot container volumes & state LiveTeams already trust Plakar.
Join us on Discord"Very efficient, free, open-source, easy as hell to integrate."
"The first big disruption to the backup / archiving world in a long time. Open source & built in Europe. Must have a look for SREs / sysadmins."
"Plakar deduplicates and encrypts at the source so your cloud provider never sees your keys. And the snapshot model has no chain dependencies. You can delete any backup without breaking the others."
"Restores 1 million items in 3 minutes."
"Plakar looks pretty slick, especially that Proxmox integration."
"Two little French geniuses just released a solution that's going to make you love backups."
Community powered
There must be a way to ensure backups remain accessible 10 or 30 years from now. Built with transparency, tested by the community and trusted for critical infrastructure.
Open source
Plakar is built by a global community of engaged developers, SREs, sysadmins and platform engineers.
Write your own plugins, contribute to the codebase or just help us.
Every chunk, accounted for. Voilà.
Nestor is Plakar's mascot, a chipmunk obsessed with chunks. His companion? Chunky Ptarson, a nut named after the .ptar format he calls home. Nestor carries the snapshots. Chunky is the snapshot.
"You're just… leaving your snapshots out there? Unverified?! Mon dieu. You have to protect your nuts!"
— Nestor, every time someone says their backups "just work"Stay ahead with Plakar updates
Get exclusive insights, new features and expert tips on backup, storage and immutability.
No spam, ever.
Sign up with your email
We'll send you major product updates, no noise.
Back up anything. Prove it's recoverable.
The complete engine, free and open source. Organization-wide visibility when you need it. Your keys, your data, your recovery path.
Plakar Blog
Engineers thoughts, release notes and data protection latest news
Plakar v1.1.3 is here
After a long beta and three release candidates, plakar v1.1.3 is finally out. It is the biggest release we have shipped so far: faster restores, much lower memory usage, a smaller cache footprint, a brand new terminal UI, better mounting, a new package manager, simpler integration interfaces, and a far more reliable execution model now that the agent is gone and cached has taken its place. It is stable, fully backward compatible, and ready for production.
Read more
Plakar v1.1.0-beta: the foundation for what’s next
plakar v1.1.0-beta marks a major step forward, with significant performance gains, architectural simplifications, and powerful new user-facing features. From faster backups and restores to better mounting, cleaner integrations, and a more reliable execution model, this release lays solid foundations for what comes next. The beta is stable, backward compatible, and ready to be explored.
Read more
Plakar joins the Linux Foundation and Cloud Native Computing Foundation
We are proud to announce that Plakar has officially joined the Linux Foundation and the CNCF as a member, marking a pivotal step in establishing an Open Standard for Resilience.
Read more