System Requirements

Meridian VMS runs on Ubuntu and supports deployments from a single Raspberry Pi to multi-server enterprise setups.

Supported Operating Systems

OS	Status
Ubuntu 24.04 LTS	Recommended
Ubuntu 22.04 LTS	Supported

Note

Ubuntu 24.04 LTS is recommended for new installations. Ubuntu 22.04 LTS is supported — the installer will automatically configure any additional dependencies needed.

Deployment Options

Deployment	Use Case	Cameras
Raspberry Pi 5	Home / small office	1-10 cameras
Single Server	Small to medium business	10-50 cameras
Multi-Server	Enterprise / large sites	50-500+ cameras

Management Server

The management server hosts the web interface, API, database, and coordinates all recording servers. For small deployments, it can run on the same machine as the recording server.

Hardware

Component	Minimum	Recommended
CPU	4 cores	8+ cores
RAM	4 GB	16 GB
Storage	50 GB SSD	100 GB SSD
Network	1 Gbps	1 Gbps

Note

The management server does not store video recordings. Storage is for the OS, database, and application. For 300+ cameras, increase RAM to 16GB+.

Recording Server

Each recording server manages video ingestion and storage for its assigned cameras. You can deploy multiple recording servers to distribute load.

Hardware

Component	Minimum	Recommended
CPU	4 cores	8+ cores
RAM	8 GB	16 GB
Storage (OS)	50 GB SSD	50 GB SSD
Storage (recordings)	1 TB HDD/SSD	Scaled to camera count and retention
Network	1 Gbps	10 Gbps for 100+ cameras

Raspberry Pi

Meridian VMS runs on Raspberry Pi 5 for home and small office deployments. Both management and recording run on a single Pi.

Supported Models

Model	RAM	Cameras	AI Detection
Raspberry Pi 5 (8GB)	8 GB	Up to 10	Hailo-8L (HAT) or Coral USB
Raspberry Pi 5 (4GB)	4 GB	Up to 5	Hailo-8L (HAT) or Coral USB
Raspberry Pi 4 (4GB)	4 GB	Up to 4	Coral USB only
Raspberry Pi 4 (8GB)	8 GB	Up to 6	Coral USB only

Note

The Pi 5 is recommended for new installations — its faster CPU and PCIe bus provide better performance, especially with AI detection via the Hailo-8L HAT. The Pi 4 works well for basic recording with a few cameras but does not support the Hailo-8L (no PCIe).

Pi Storage

• OS: 32GB+ microSD card or USB SSD (recommended)
• Recordings: USB SSD or NVMe via HAT — microSD is too slow for recording
• Recommended: Pimoroni NVMe Base Duo — supports an NVMe drive AND the M.2 Hailo-8L accelerator on the same board, so you get fast storage and AI detection in one HAT

Pi Network

• Use Ethernet (Gigabit) — Wi-Fi is not reliable enough for multi-camera recording
• PoE HAT available for clean single-cable installation

AI Detection Accelerators

AI object detection requires a hardware accelerator. Meridian VMS auto-detects the available hardware and selects the best backend.

NVIDIA GPU (recommended for servers)

Component	Minimum	Recommended
GPU	NVIDIA with 4 GB VRAM	NVIDIA Tesla T4 (16 GB) or better
CUDA	11.8+	12.x
Driver	525+	Latest stable

Best for high camera counts (30+ cameras per GPU). Supports all detection features including custom model training.

Google Coral TPU

Variant	Interface	Performance
Coral USB Accelerator	USB 3.0	~5-8 cameras
Coral M.2 / Mini PCIe	PCIe	~10-15 cameras

Works on Raspberry Pi and x86 servers. Lower power consumption than GPU. No custom model training — uses pre-built models.

Hailo-8L (Raspberry Pi)

Variant	Interface	Performance
Hailo-8L M.2	PCIe (Pi 5 HAT)	~8-10 cameras

Purpose-built for edge AI on Raspberry Pi 5. Available as an M.2 module on the official Raspberry Pi AI HAT. Excellent power efficiency.

No Accelerator

Meridian VMS works without any AI accelerator — cameras record and stream normally. You can add a detection accelerator later without reinstalling.

Detection Accelerator Comparison

Feature	NVIDIA GPU	Coral TPU	Hailo-8L	CPU (no accel)
Cameras	30-300+	5-15	8-10	1-3
Custom training	Yes	No	No	No
LPR/ANPR	Yes	Yes	Yes	Slow
Power draw	70-250W	2-4W	3W	—
Cost	$300-2000+	$60-80	$70	Free
Pi compatible	No	USB only	Yes (HAT)	Yes

Recording Storage Sizing

Camera Resolution	Bitrate (typical)	Storage per Camera per Day
1080p (2 MP)	4 Mbps	~43 GB
4 MP	6 Mbps	~65 GB
4K (8 MP)	10 Mbps	~108 GB

Storage calculation

Total storage = (cameras) x (GB per camera per day) x (retention days)

Example: 10 cameras at 4 MP with 14-day retention = 10 x 65 GB x 14 = 9.1 TB

Network Requirements

Ports

From	To	Port	Protocol	Purpose
Browser	Management server	8000	TCP/HTTP	Web UI and API
Recording server	Management server	8000	TCP/HTTP	Registration, heartbeat
Browser	Recording server	8554	TCP/RTSP	Live streaming (local)
Browser	Recording server	8555	UDP	WebRTC P2P (remote access)
Both servers	MQTT broker	1883	TCP	Real-time messaging

Camera Network

• All cameras must be reachable from their recording server via RTSP (typically port 554)
• ONVIF discovery requires cameras and the management server to be on the same subnet
• A dedicated camera VLAN is recommended to isolate surveillance traffic

Supported Cameras

Meridian VMS supports any IP camera with an RTSP stream:

• Required: RTSP streaming (H.264 or H.265)
• Recommended: ONVIF Profile S for autodiscovery and PTZ control
• Optional: ONVIF events for camera-native motion detection

Note

Meridian VMS never re-encodes video. The camera must provide a properly formatted H.264 or H.265 stream.

Software Requirements

The web-based installer handles all software installation automatically. Ubuntu 24.04 LTS is recommended. Ubuntu 22.04 LTS is also supported — the installer handles any additional setup needed.

Pre-Installation Checklist

• Ubuntu 24.04 LTS (recommended) or 22.04 LTS installed
• Network connectivity established
• Recording storage mounted and accessible
• IP cameras powered on and reachable via RTSP
• Internet access available (for installer downloads)