Storage · Hosted in Germany
Ceph Cluster
In many environments, storage is the hidden single point of failure. Ceph solves the problem at its root: distributed, self-healing storage across all nodes – without a central storage system that could fail.
Fundamentals
What a Ceph cluster is.
Ceph is a distributed storage system: instead of keeping data on one central storage device, it spreads every data block in multiple copies across the disks of all cluster nodes. If a disk or an entire node fails, the data remains fully available – and Ceph restores the target redundancy on its own, without manual intervention.
In our Managed Proxmox Clusters, Ceph is integrated directly into the virtualization layer: the virtual machines live on the distributed storage and can therefore start on any node – the foundation for automatic failover and live migration.
Characteristics
What Ceph delivers in an HA cluster.
- No single point of failure: no central storage system whose failure stops everything.
- Multiple replicas: every data block is stored in several copies on different nodes.
- Self-healing: after a disk or node failure, Ceph restores redundancy automatically.
- Scalable: more capacity or throughput means additional disks or nodes – without a migration.
- Dedicated storage network: replication traffic runs separately from user traffic, on NVMe systems if required.
- Integrated into Proxmox VE: management, monitoring, and maintenance from a single source – as part of managed operations.
Trade-offs
Ceph or ZFS replication?
Ceph plays to its strengths in clusters with three or more nodes and growing data volumes. For more compact setups, ZFS with replication is often the more sober choice: fewer moving parts, very robust and proven technology, but replication at intervals rather than synchronous.
What fits your application is decided by the load profile – IO behavior, data volume, RPO target. That is exactly what the design phase is for: you get a recommendation with reasoning, not technology for its own sake.
And regardless of the storage concept: replicated storage is no substitute for a backup. Separate backup systems with restore tests are always part of our package.
FAQ
Common questions about the Ceph cluster.
When is Ceph worth it?
From three nodes upward – below that, Ceph cannot distribute its redundancy in a meaningful way. Rule of thumb: the more nodes and the more dynamic the growth, the more Ceph plays to its strengths. For smaller setups we usually recommend ZFS replication.
What happens when a disk fails?
Nothing your users will notice: the data exists in further copies on other disks and nodes. Ceph restores full redundancy automatically, and our operations team replaces the faulty disk – documented in a follow-up report.
How many copies of the data are kept?
The standard is three replicas across different nodes – the cluster survives the simultaneous failure of two disks or an entire node without data loss. We define the replication scheme in the concept.
Does Ceph replace backups?
No. Ceph protects against hardware failure, but it also replicates mistakes – an accidental deletion is immediately on every copy. Separate backups with restore tests are therefore always part of the setup.
Contact
Storage without sleepless nights.
Briefly describe your application and data volume. You will receive a storage recommendation within one business day, free of charge and without obligation.
Request a project