Implementation of agent interaction protocol within cloud infrastructure in geographically distributed data centers

A cloud system for geographically distributed data centers is presented. An approach is based on the principle of multi-agent structure (microservices). A special interaction protocol was developed for agents’ communication operating in asynchronous mode. The asynchronous behavior of the agent interaction system is achieved through the use of a specially developed algorithm. The program that implements the proposed algorithm is written in the Python programming language. This solution uses relational databases and queuing systems. A relational database stores requests and responses from the agents. A message broker is necessary for exchanging YAML messages with identifiers of these requests and responses. The developed software was tested on a prototype of a scalable geographically distributed data center. An original technical solution was obtained that successfully passed a series of tests and was implemented within existing cloud infrastructure. Features of applying the RabbitMQ queuing system and PostgreSQL database management system in cluster mode with traffic encryption are indicated. The use of the developed model appears to be promising oriented for highly loaded distributed systems.