Rubrik · Infrastructure · YoctoIT tech page

Rubrik Cluster (CDM)

The scale-out on-premises appliance: nodes that grow horizontally, an immutable filesystem and global deduplication — the Rubrik data plane inside your data center.

FOCUS · THE ON-PREM DATA PLANEReal scale-out: start with 3 nodes and grow brick by brick, no forklift upgrades
Request the Free Assessment Learn more
YoctoIT material for clients and partners · Rubrik and the other products mentioned are trademarks of Rubrik, Inc.
01 · What it is

Rubrik Cluster (CDM), made clear.

Rubrik CDM (Cloud Data Management) is the cluster running backup and recovery inside your perimeter: r6000-series branded appliances or certified Dell, HPE, Lenovo and Cisco hardware. Every node adds capacity and power: no media servers, no external storage, no single point of failure — and the append-only filesystem makes the copies untouchable.

3+
nodes to start: you grow one brick at a time
0
media servers or external storage to design and maintain
r6000
the current appliance series — or certified iron from the big four
Rubrik Cluster (CDM)
OFFICIAL BRANDING RUBRIK
02 · How to use it well

The things that make the difference.

The cluster

VMware / Hyper-V / physical / DB productionwhat needs protecting
Node 1
Node 2
Node N
scale-out · global dedup · erasure coding
Atlas append-only filesystemthe immutable copies
Replica & archiveto a second site or object storage
On-prem backup without weak points

Honest sizing

Real capacity, change rate and retention: the right cluster, not the price-list one.

Zero-downtime upgrades

Nodes update in rotation: the backup service never stops.

Replication where needed

To a second cluster, the cloud or both: the geography of resilience.

Measured restores

Live Mount and instant recovery: RTO in minutes, verified in periodic tests.

03 · In depth

Scale-out and immutability: the cluster architecture

Every CDM node brings compute, storage and network; the distributed Atlas filesystem writes append-only with erasure coding across nodes: losing disks or an entire node doesn't stop the cluster; deduplication is global; Live Mount boots VMs directly from backup for recovery in minutes and copy-free tests; REST APIs integrate the cluster into vRO, Ansible and our NOC runbooks.

  • Atlas append-only — the filesystem that never overwrites: native immutability, not bolted on
  • Erasure coding — protection from disk/node failures with real efficiency, not 3x replicas
  • Live Mount — the VM restarts mounted from backup: RTO in minutes, then storage vMotion
  • Global dedup — across all nodes and workloads: the TBs you never buy
  • Certified hardware — r6000 or Dell/HPE/Lenovo/Cisco: choice without iron lock-in
  • Full REST APIs — automation and integration: the cluster is a citizen of your stack
04 · Numbers and lifecycle

The numbers that matter.

N+1
cluster fault tolerance: disks and nodes
min
recovery times with Live Mount
4
certified hardware vendors beyond the r6000 appliances
1 hop
integrated replication towards DR and cloud
On-prem backup shouldn't be a storage project: with CDM it's a cluster that grows with you — and we run it 24/7.
05 · Use cases

Where it really pays off.

Enterprise data center

Local protection with RTO in minutes.

Legacy backup refresh

Goodbye fragile media servers and tape libraries: one cluster.

Two-site DR

Clusters in replication: the second site always ready.

A well-sized cluster is worth more than a thousand slides: let's do the math on your real change rate.