Microservice based architecture for the genetic algorithm

Abstract

Microservice architecture is becoming more popular and more frequently used, mainly because of its numerous advantages over monolith approach. Namely, the developed systems nowadays need more agile distribution of the processing power, and due to their size, a way to deploy, maintain and test individual components separately. Each algorithm/software composed of individual and independently executable parts that do not share many parameters are good candidates for a solution based on a microservice architecture. This paper presents a microservice approach when building an architecture for the genetic algorithm. We identified eight independent parts of the genetic algorithm, and each one is represented as a microservice. This design leads to a solution that has low coupling and high cohesion. The advantages of this approach include distributing the computations on more physical locations, and furthermore, scaling only the parts of the system which require more performance (which means need large processing power or are frequently executed).