Notice

This document is for a development version of Ceph.

Minimum Hardware per Daemon

The smallest configuration that each daemon runs with. Cores means threads when hyperthreading is enabled. Production clusters need more; see Hardware Recommendations.

Process

Criteria

Minimum (recommended where shown)

ceph-osd

Processor

1 thread minimum, 3 recommended per HDD OSD; 4 minimum, 6 recommended per NVMe SSD OSD.

RAM

>= 4 GiB per OSD; see CPU and Memory Sizing

Storage Drives

1x storage drive per OSD in most cases. PCIe Gen 4+ SSDs larger than 30 TB may benefit from being split into two or more OSDs.

DB/WAL offload (optional)

1x SSD partition per HDD OSD 4-5x HDD OSDs per DB/WAL SAS or SATA SSD <= 15 HDD OSDs per DB/WAL NVMe SSD

Network

1 Gb/s minimum; 10 Gb/s recommended (Network Sizing)

ceph-mon

Processor

2 cores minimum

RAM

>= 5 GB per daemon (large / production clusters need more)

Storage

100 GB per daemon, SSD strongly urged

Network

1 Gb/s minimum; 10 Gb/s recommended (Network Sizing)

ceph-mds

Processor

2 cores minimum, higher freq is better than more cores

RAM

>= 8 GiB per daemon

Network

1 Gb/s minimum; 10 Gb/s recommended (Network Sizing)

Thread counts are before replication and vary with CPU and drive model, erasure coding, and compression; ARM CPUs may need more cores. Benchmark.

Warning

Use enterprise-quality storage media for production workloads.

Tip

Use a separate drive for the OS. If a node has only one drive, put the OSD on its own partition. See Storage Devices.

Additional Resources

Brought to you by the Ceph Foundation

The Ceph Documentation is a community resource funded and hosted by the non-profit Ceph Foundation. If you would like to support this and our other efforts, please consider joining now.