System Architecture

Three-Engine + AI Agent Full-Stack Operations Platform

Architecture Overview

OpsDash uses a decoupled frontend/backend architecture. The frontend is built with Vue 3.5 + TypeScript 5.5 for a modern interactive experience, while the backend leverages FastAPI 0.115 async framework for high-performance API services. Monitoring is powered by three engines: Zabbix 7.0 covers traditional IT infrastructure, EMQX 5.8 supports IoT and industrial device connectivity, and VictoriaMetrics 1.106 serves as the unified time-series storage engine aggregating metrics across the platform. The MCP Server exposes 189 standardized tools, enabling AI Agents to perform operations tasks using natural language. The data layer relies on PostgreSQL 16 for persistent business state and Redis for caching and real-time message delivery.

💻 Users / Browser 🤖 AI Agents (Claude / Cursor) Vue 3.5 + TypeScript Ant Design Vue · 72 Pages · WebSocket MCP Server :8002 189 Tools · 11 Resources · 14 Prompts FastAPI 0.115 Async Backend 56 Routers · 404 Endpoints · 62 ORM Models · 47 Service Modules · RBAC 5 Roles Zabbix Server 7.0 Traditional IT: SNMP / ICMP / SSH / IPMI EMQX 5.8 IoT / Industrial: MQTT / CoAP / Modbus VictoriaMetrics 1.106 Unified Time-Series · PromQL Query PostgreSQL 16 Business Data · 62 Tables · Async Pool Redis Cache · PubSub · WebSocket Push Monitored Devices 🖥 Servers 🌐 Network 📡 IoT Sensors 🏭 Industrial HTTPS SSE / stdio REST API WebSocket HTTP API Zabbix API HTTP API + Webhook PromQL asyncpg PubSub Time-Series Alert Webhook SNMP/ICMP MQTT Access Presentation Service Engines Data

Three-Engine Comparison

Three engines, each with a distinct role, unified under the OpsDash platform

Dimension Zabbix 7.0 EMQX 5.8 VictoriaMetrics
Role Traditional IT infrastructure IoT / Industrial devices Unified time-series storage
Protocols SNMP / ICMP / SSH / IPMI / WMI / JMX / HTTP / TCP MQTT / CoAP / Modbus PromQL
Device Types Servers / Network / Security / Database / Storage IoT sensors / Gateways / PLCs All engine data aggregation
Data Model Item → Trigger → Action Topic → Rule Engine Metric + Label
Scaling Proxy hierarchy Clustering Single-node high performance

Deployment Options

Flexible deployment scales, smoothly upgrade from evaluation to large-scale production

Evaluation

Try all features quickly

  • 2 CPU Cores / 4 GB RAM
  • 50 GB SSD Disk
  • Single-node Docker Compose
  • ≤ 20 Devices (Free tier)
  • 5-minute installation

7 containers use ~2 GB RAM total. VM 90-day data ~2 GB disk. VM or bare metal.

Production

Mid-size enterprise deployment

  • 8 CPU Cores / 16 GB RAM
  • 200 GB SSD Disk
  • 50 — 200 Devices
  • 90-day data retention
  • Optional Zabbix Proxy tiers

VM 90-day data ~10-50 GB, PG data ~200 MB-800 MB. Gigabit network recommended.

Large Scale

Enterprise distributed deployment

  • 16 CPU Cores / 32 GB RAM
  • 500 GB+ SSD/NVMe Disk
  • 200 — 1000+ Devices
  • Zabbix Proxy tiers (required)
  • PG replication + Redis Sentinel recommended

VM 90-day data ~100-400 GB. Each Proxy node: 2 cores 4 GB, handles 50-200 devices. 10GbE recommended.

Data Flow

Complete data pipeline from device collection to AI-powered operations

  1. Device Collection — Devices report data to their respective engines via Agent / SNMP / MQTT
  2. Engine Aggregation — Time-series metrics are written to VictoriaMetrics; business state is stored in PostgreSQL
  3. API Distribution — FastAPI backend unifies aggregated queries and pushes data to the frontend Dashboard
  4. AI Integration — MCP Server exposes 189 tools, enabling AI Agents to perform operations tasks using natural language
SOURCES ENGINES DATA + BACKEND CONSUMERS Servers Agent / SNMP Network SNMP / ICMP IoT Sensors MQTT / CoAP Zabbix Server Webhook + History Export to VM EMQX Rule Engine HTTP Auth + Webhook VictoriaMetrics Time-series Storage PostgreSQL Business Data Redis Cache + PubSub OpsDash Backend FastAPI 56 Routers / 404 APIs Alert Engine + WS Frontend Vue 3 Dashboard 72 Pages / Real-time MCP Server 189 Tools AI Agent Claude / Cursor SNMP Export Webhook MQTT Telemetry GPS+Auth REST API PromQL SQL WebSocket MCP REST 2-Step Approval Legend IT Monitoring IoT Data Query & Display AI Agent

Tech Stack

Modern full-stack technology choices, balancing performance and development efficiency

Frontend

Vue 3.5 TypeScript 5.5 Ant Design Vue 4.2 ECharts 5.5 AntV G6 5.0 Leaflet xterm.js Monaco Editor

Backend

FastAPI 0.115 SQLAlchemy 2.0 Alembic Pydantic v2 httpx asyncssh PostgreSQL 16 Redis

Start Monitoring in 5 Minutes

One-click Docker deployment, free for up to 20 devices

Get Started