Red Hat Edge Manager system requirements

The Red Hat Edge Manager control plane on Red Hat OpenShift Container Platform requires minimum cluster CPU, memory, and persistent storage that scale with the number of managed devices.

The following table shows minimum recommended cluster CPU, memory, and persistent storage for the Red Hat Edge Manager deployment based on the number of managed devices. These values were established through benchmarking and represent the resources needed for the server workload, not the managed edge devices themselves.

Table 1. Minimum recommended cluster resources
Deployment size Managed devices vCPU (cores) vRAM (GB) vDisk (GB)

Small

100

4

4

70

Medium

1,000

4

8

70

Large

10,000

4

16

70

Super Large

100,000

8

40

120
Note

These figures are an approximate starting baseline, not a guarantee. Actual resource usage depends on polling intervals, fleet template complexity, and observability stack configuration. Validate against your own workload and adjust as needed. Add approximately 3 GB disk for each additional service component (image builder, observability stack).

Per-component persistent storage
Copy link

The persistent storage listed in the cluster resources table is distributed across the following components:

Table 2. Persistent storage by component
Component Minimum capacity Notes

Built-in PostgreSQL database

60 GiB

Stores device inventory, fleet templates, and system state. Configurable through the db.builtin.storage.size Helm value.

Alertmanager

2 GiB

Stores alert state and silences. Fixed capacity.

Additional storage is required for optional components such as the image builder worker and the observability stack.

Important

Do not use NFS for the PostgreSQL database PersistentVolume (PV). NFS does not support chown operations that the PostgreSQL container requires during startup to set data directory ownership via the pod fsGroup security context. Use block storage, local volumes, or another non-NFS storage backend for the database PV.

Static storage provisioning
Copy link

On clusters without dynamic storage provisioning, you must pre-provision PersistentVolumes and configure static PV binding through Helm chart values before installation. For details, see Configuring static PV binding for clusters without dynamic storage provisioning.