Methodology for automating enterprise business processes based on API integration
In the context of rapid digitalisation and a high level of heterogeneity of contemporary corporate ecosystems, the problem of fragmentation of the information space, and the emergence of isolated data warehouses becomes a critical barrier to business efficiency. The purpose of the study was to develop and substantiate a comprehensive methodology for automating business processes of an enterprise based on the application programming interface – integration of the application programming interface, combining the choice of optimal architectural patterns with mechanisms for secure data exchange. To achieve this goal, the researchers applied system analysis of architectural styles, prototyping of software gateways in Python, and computer modelling of load scenarios in an isolated virtual environment using Docker containers and the Locust tool. The study involved a comparative analysis of the Representational State Transfer, Simple Object Access Protocol, and Graph Query Language protocols in terms of performance and network load, which revealed significant traffic overhead and delays of up to 185 ms when using outdated standards. A hybrid integration methodology was proposed that combines the Representation State Transfer architecture for mobile clients and Graph Query Language for web interfaces, which reduced the amount of transmitted data by 73% and reduced the system response time to 65 ms. A secure data exchange algorithm has been developed based on the use of an intermediate layer (middleware) and a security gateway (API Gateway), which implement two-level request validation and centralised access control to neutralise the risks of unauthorised interference. The expediency of using the "Strangler Fig" migration pattern was substantiated, which allowed for a gradual transition from monolithic enterprise resource planning systems to microservice architecture through a specialised Python gateway without stopping operational activities. It has been experimentally proven that the introduction of asynchronous message queues allows the system to maintain stable operation within 320 ms even at peak loads of up to 5,000 requests per second, in contrast to monolithic solutions that demonstrate exponential performance degradation