Skip to content

Prerequisites

Overview

Before installing Authenta On-Prem, ensure that your system meets the hardware, software, and network prerequisites listed below.
These requirements are designed to guarantee stable performance, successful image pulls, and full offline operation after setup.

1. System Requirements

Authenta supports both CPU and GPU environments.
Hardware specifications vary based on the volume of media processed and desired inference speed.

✅ Minimum Recommended Hardware

ComponentRequirementNotes
CPU8-core Intel Xeon / AMD EPYC (or higher)Required for CPU-only inference
GPU (optional)NVIDIA GPU with ≥8GB VRAM (Tesla / RTX series)Required for GPU-accelerated mode
Memory (RAM)16 GB minimum (32 GB recommended)For concurrent processing and caching
Storage100 GB free disk spaceFor Docker images, logs, and temporary media files
NetworkTemporary internet accessOnly required once to pull images from AWS ECR
OSUbuntu 20.04+ / CentOS 8+ / Amazon Linux 264-bit Linux recommended

💡 Tip: For production deployments or large-scale workloads, allocate dedicated GPU resources and SSD storage for improved throughput.

2. Software Requirements

Authenta relies on standard container and CLI tooling.
Ensure the following software is installed on your host system:

SoftwarePurposeInstallation Reference
Docker Engine (v20.10 or newer)Runs all Authenta containers.Install Docker Engine →
Docker Compose (v2.5 or newer)Orchestrates Authenta services.Install Docker Compose →
AWS CLI (v2 or newer)Required for authentication to Authenta’s private ECR.Install AWS CLI v2 →
Node.js (v16+ optional)Runs the sample demo application.Node.js Downloads →

⚙️ Verify your installations:

docker --version
docker compose version
aws --version

3. GPU Environment (Optional)

If you plan to use GPU acceleration (ml-task-runner-gpu profile), you must install and configure the NVIDIA Container Toolkit and related drivers.

Required Components

DependencyDescription
NVIDIA DriverMust match your GPU model and CUDA version. The driver requirement is described in NVIDIA’s installation guide — follow it closely.
NVIDIA Container ToolkitEnables Docker containers to access and utilize GPU resources. Must also be configured for Docker runtime integration.

Follow NVIDIA’s official setup and configuration guide:
👉 Install and Configure NVIDIA Container Toolkit →

⚠️ Important:
The toolkit installation guide includes a step to enable GPU support in Docker (by configuring the NVIDIA runtime in Docker’s settings). Make sure to complete this step — Authenta’s GPU containers require Docker to recognize the NVIDIA runtime. Skipping this configuration will cause GPU-enabled containers to fail at startup.

Once installation and configuration are complete, you can validate your GPU environment using NVIDIA’s own sample workload:
👉 Run NVIDIA Sample Workload →

Authenta does not require any additional verification beyond the steps in NVIDIA’s documentation.

4. Network & Access Requirements

Authenta operates fully offline after setup, but you'll need limited internet access initially to pull Docker images from our private AWS registry.

Required During Setup

ItemDescription
Outbound Internet AccessNeeded only once to pull container images. Can be disabled afterward
AWS ECR CredentialsProvided by Authenta. Used for secure login and image pull
PortsInternal communication only (RabbitMQ: 5672 / 15672 for dashboard)
Firewall RulesAllow local container-to-container communication over the Docker network

⚠️ After installation, you may safely disable outbound internet and lock down ports for air-gapped operation.

5. File System & Storage Setup

Authenta requires a shared host directory that is mounted into containers for data exchange.

PathPurpose
/opt/authenta/dataStores input files and inference outputs
/opt/authenta/logs(Optional) Capture logs for auditing or monitoring

Ensure that:

  • The directory exists before starting the containers
  • Proper permissions are assigned (e.g., chmod -R 775 /opt/authenta/data)
  • Sufficient disk space is available for temporary media storage

6. Access Permissions

To deploy Authenta, the executing user must have:

  • Docker permissions (part of the docker group)
  • Read/Write access to shared data and log directories
  • Network access (for initial setup only)
  • Authenta Registry Access: Authenta will provide your team with the AWS Account ID and temporary ECR credentials needed to pull images from Authenta’s private registry.

7. Optional Tools (for Debugging or Integration)

While not mandatory, the following tools are useful during setup or testing:

ToolPurpose
Postman / curlTest message publishing to RabbitMQ or APIs
Docker Desktop / PortainerVisual management of containers and logs
jqJSON parsing utility for log inspection

Summary

CategoryRequirement
Hardware8-core CPU, 16–32 GB RAM, 100 GB storage
GPU (optional)NVIDIA GPU with drivers + container toolkit
SoftwareDocker, Docker Compose, AWS CLI
NetworkOne-time internet access for ECR image pull
PermissionsDocker group membership + directory access
OS64-bit Linux (Ubuntu 20.04+, CentOS 8+, Amazon Linux 2)