MQTT is a great data delivering service protocol - lightweight and secure via TLS encryption and with user login in communication. The architecture of MQTT is easy to provide with Docker containers or Kubernetes Pods. It is important for AI services to have a data infrastructure and to get just the data it needs to proccess in tasks.

 

Data Stores and Data Lakes can help to hold and store data in different time ranges. That saves storage and money for licenses, servers and other hardware.

 

You can save a lot of data transportation and costs, especially with "on-edge" solutions and Edge-Computing. Nowadays Edge-AI is provided by NVIDIA, GOOGLE and INTEL the most with hardware and software. This is the best constallation. All fits well together.

If you have new tasks for AI or ML services, then start with the right MQTT and data architecture and infrastructure first!

 

PostgreSQL and EDB are great tools to combine them with MQTT Brokers from HiveMQ or Cedalo. MQTT v5 and SparkPlug B are well designed for the industry and we are very happy with the stability and comfort of those tools.