Hardware requirements
These hardware requirements are based on internal testing at Puppet and are provided as minimum guidelines to help you determine your hardware needs.
Your configuration and code base can significantly affect performance. Use PE tuning and metrics tools to further customize and refine your installation.
If possible, address performance limitations by maximizing your hardware first, then scaling up to the next size architecture as needed. It's often easier to upgrade your hardware than to add additional infrastructure nodes.
Hardware requirements for standard installations
These are the minimum hardware requirements for the primary server in a standard architecture with up to 2,500 nodes.
| Node volume | Cores | RAM | /opt/ | /var/ |
|---|---|---|---|---|
| Trial use | 2 | 8 GB | 20 GB | 24 GB |
| 11–100 | 6 | 10 GB | 50 GB | 24 GB |
| 101–500 | 8 | 12 GB | 50 GB | 24 GB |
| 501–1,000 | 10 | 16 GB | 50 GB | 24 GB |
| 1,000–2,500 | 12 | 24 GB | 50 GB | 24 GB |
- Trial mode: Although the m5.large instance type is sufficient for trial use, it is not supported. A minimum of four cores is required for production workloads.
- /opt/ storage requirements: The database should not exceed 50% of /opt/ to allow for future upgrades.
- /var/ storage requirements: There are roughly 20 log files stored in /var/ which are limited in size to 1 GB each. We recommend allocating 24 GB to avoid issues, however log retention settings generally prevent reaching the maximum capacity.
Hardware requirements for large installations
These are the minimum hardware requirements for the primary server and compilers in a large architecture with 2,500–20,000 nodes.
Each compiler increases capacity by approximately 1,500–3,000 nodes, until you exhaust the capacity of PuppetDB or the console, which run on the primary server.
| Node volume | Node | Cores | RAM | /opt/ | /var/ | EC2 |
|---|---|---|---|---|---|---|
| 2,500–20,000 | Primary node | 16 | 32 GB | 150 GB | 10 GB | c5.4xlarge |
| Each compiler (1,500 - 3,000 nodes) | 6 | 12 GB | 30 GB | 2 GB | m5.xlarge |
Hardware requirements for extra-large installations
These are the minimum hardware requirements for the primary server, compilers, and PE-PostgreSQL nodes in an extra-large architecture with 20,000+ nodes.
| Node volume | Node | Cores | RAM | /opt/ | /var/ | EC2 |
|---|---|---|---|---|---|---|
| 20,000+ | Primary node | 16 | 32 GB | 150 GB | 10 GB | c5.4xlarge |
| Each compiler (1,500 - 3,000 nodes) | 6 | 12 GB | 30 GB | 2 GB | m5.xlarge | |
| PE-PostgreSQL node | 16 | 128 GB | 300 GB | 4 GB | r5.4xlarge |
If you manage more than 20,000 nodes, contact your technical account manager or Puppet Professional Services to talk about optimizing your setup for your specific requirements.
Hardware requirements for cloud deployments
Ensure that your primary server meets the minimum hardware requirements for cloud deployments. Cloud deployments use a standard architecture with up to 2,500 nodes.
| Node volume | Amazon Web Services (AWS) | Azure |
|---|---|---|
| Trial use | m5.large | D2 v4 |
| 11–100 | c5.2xlarge | F8s v2 |
| 101–500 | c5.2xlarge | F8s v2 |
| 501–1,000 | c5.2xlarge | F8s v2 |
| 1,000–2,500 | c5.4xlarge | F16s v2 |
Azure requirements are not currently tested by Puppet, but are presented here as best guidance based on comparable AWS testing.