Adds three new modules that let business code call center services
through protocol-neutral interfaces instead of @GrpcClient stubs:
- dc3-common-facade-api: Device/Driver/Point + Tenant/User/UserLogin/Token
facade interfaces plus facade-owned BO/Query/Page types
- dc3-common-facade-local: in-process implementations backed by the
auth/manager services; zero serialization cost in single deployments
- dc3-common-facade-grpc: gRPC implementations holding BlockingStubs;
default mode for distributed deployments
Implementation selection is driven by dc3.facade.mode={local|grpc} via
@ConditionalOnProperty, with grpc as the matchIfMissing default. Wire-up
for dc3-center-single (dc3.facade.mode=local + facade-local dependency)
is included so the single-deployment mode is runnable out of the box.
English | 中文 | 日本語 | Tiếng Việt
IoT DC3 is a fully open-source, distributed Internet of Things (IoT) platform built on Spring Cloud.
It accelerates IoT solution delivery and simplifies full-lifecycle device management with a comprehensive architecture for robust, production-ready IoT systems.
It is AI-ready, enabling seamless integration of intelligent connectivity, automation, and data-driven operations.
All components and code are open-source, ensuring transparency, flexibility, and community-driven innovation.
1 Architecture
The architecture is designed for end-to-end IoT capabilities across device connectivity, data services, operational management, and extensible application integration.
- Driver Layer: Provides SDKs for rapid driver development and seamless connectivity to physical devices through standard or proprietary protocols. This layer handles southbound data acquisition and command execution;
- Data Layer: Supports reliable collection, storage, and retrieval of device data, exposing robust interfaces for real-time and historical data services;
- Management Layer: Serves as the core hub for distributed microservice collaboration, including service registration, device/driver management, command orchestration, and centralized configuration governance;
- Application Layer: Enables data openness, scheduling, alarms, messaging, logging, third-party integrations, and AI-enhanced automation scenarios.
2 Objectives
- Scalability: Supports horizontal scaling with Spring Cloud for distributed, high-throughput IoT workloads;
- Resilience: Minimizes single-point-of-failure risk with interchangeable service nodes and fault-tolerant design;
- Performance: Handles large-scale device access and telemetry workloads for demanding IoT scenarios;
- Extensibility: Accelerates integration of new protocols and custom drivers through SDK and service registration;
- Deployment Flexibility: Runs across private cloud, public cloud, and edge environments with Java compatibility;
- Operational Efficiency: Streamlines onboarding, registration, and permission validation for devices and services;
- Security and Multi-Tenancy: Enforces encrypted communication, namespace isolation, and tenant-level separation;
- Cloud-Native Delivery: Optimized for Kubernetes and containerized with Docker for consistent deployments;
- AI-Ready Evolution: Enables integration of intelligent automation and data-driven operational workflows.
3 Development
3.1 Startup Dependencies
Choose one
This base stack starts PostgreSQL and RabbitMQ. If you need a database SQL script, connect directly to the started database in the container for export.
# Global access with standard container registry service
podman compose -f dc3/docker-compose-db.yml up -d
# Optimized registry service for users in mainland China
podman compose -f dc3/docker-compose-db-aliyun.yml up -d
Optional helper targets:
make dev-db
make dev-optional
make dev
make dev-all
Use REGISTRY=domestic when you want the mainland China image registry variants. Backward-compatible aliases REGISTRY=aliyun and REGISTRY=cn still work:
make dev-db REGISTRY=domestic
make dev-all REGISTRY=domestic
make app-all REGISTRY=aliyun
make compose-up STACK=grafana REGISTRY=cn
make compose-logs STACK=dev REGISTRY=global
Docker Compose environment overrides
Copy the example file before changing any published ports, image tags, or observability settings:
cp .env.example .env
Common overrides used by the Compose stacks in dc3/:
| Variable | Default | Purpose |
|---|---|---|
DC3_IMAGE_TAG |
2026.5 |
Shared image tag for application, database, EMQX, and observability images |
DC3_LOG_MAX_SIZE |
10M |
Maximum size per container log file |
DC3_LOG_MAX_FILE |
20 |
Number of rotated log files to keep |
DC3_BIND_HOST |
127.0.0.1 |
Host address used for published ports; set 0.0.0.0 if you need LAN access |
POSTGRES_USERNAME |
dc3 |
Username used by the Postgres health check |
POSTGRES_DB |
dc3 |
Database name used by the Postgres health check |
DC3_POSTGRES_PORT |
35432 |
Published PostgreSQL port |
DC3_RABBITMQ_TLS_PORT / DC3_RABBITMQ_PORT |
35671 / 35672 |
RabbitMQ TLS / AMQP ports |
DC3_RABBITMQ_MANAGEMENT_PORT |
15672 |
RabbitMQ management UI port |
DC3_GATEWAY_PORT |
8000 |
Gateway HTTP port |
DC3_AUTH_PORT / DC3_AUTH_GRPC_PORT |
8300 / 9300 |
Auth center HTTP and gRPC ports |
DC3_MANAGER_PORT / DC3_MANAGER_GRPC_PORT |
8400 / 9400 |
Manager center HTTP and gRPC ports |
DC3_DATA_PORT / DC3_DATA_GRPC_PORT |
8500 / 9500 |
Data center HTTP and gRPC ports |
DC3_LISTENING_VIRTUAL_DRIVER_PORT / DC3_LISTENING_VIRTUAL_GRPC_PORT |
6270 / 6271 |
Listening virtual driver ports |
DC3_EMQX_WS_PORT / DC3_EMQX_WSS_PORT |
38083 / 38084 |
EMQX WebSocket endpoints |
DC3_EMQX_MQTT_PORT / DC3_EMQX_MQTTS_PORT |
31883 / 38883 |
EMQX MQTT / MQTTS ports |
DC3_EMQX_DASHBOARD_PORT |
18083 |
EMQX dashboard port |
GF_SERVER_ROOT_URL |
http://localhost:3000 |
Grafana external root URL |
DC3_GRAFANA_PORT |
3000 |
Grafana HTTP port |
DC3_KIBANA_PORT |
5601 |
Kibana HTTP port |
DC3_ES_JAVA_OPTS |
-Xms512m -Xmx512m |
Elasticsearch JVM heap settings |
DC3_LS_JAVA_OPTS |
-Xms256m -Xmx256m |
Logstash JVM heap settings |
3.2 Preparation
source dc3/env/dev.env.sh
mvn -s .mvn/settings.xml clean package
3.3 Start Services
Start in order
# Gateway
java -jar dc3-gateway/target/dc3-gateway.jar
# Auth Center
java -jar dc3-center/dc3-center-auth/target/dc3-center-auth.jar
# Data Center
java -jar dc3-center/dc3-center-data/target/dc3-center-data.jar
# Manager Center
java -jar dc3-center/dc3-center-manager/target/dc3-center-manager.jar
# Virtual Driver
java -jar dc3-driver/dc3-driver-virtual/target/dc3-driver-virtual.jar
# Other driver: Listening Virtual Driver, Modbus TCP Driver, MQTT Driver, OPC DA Driver, OPC UA Driver, Siemens S7 Driver
4 Technology Stack
5 Contribution
- Branch Creation: Start by creating a new branch from the
mainbranch. Ensure that themainbranch is up-to-date before branching out; - Branch Naming: Follow the naming convention for the new branch:
feature/your_name/feature_description. For example:feature/pnoker/mqtt_driver; - Code and Documentation: Make your changes to the code or documentation on the new branch. Once done, commit your changes;
- Pull Request: Submit a
Pull Request(PR) to merge your changes into thedevelopbranch. Your PR will be reviewed and merged by the maintainers.
6 License
The IoT DC3 open-source platform is licensed under the AGPL 3.0 License.
