Towards Early Distribution of Container-Based Microservices in Cloud Computing Environment

Abstract

Microservices gained the popularity for use in scalable cloud application. The application based on microservices uses intensive network communication to call other microservices or to pass on the messages to hundreds of microservices running inside the cloud/edge nodes. This study addressed the early distribution of container-based microservices and proposed two distribution strategies named as Random Distribution and Design Pattern Distribution. In random distribution approach the microservices are assigned arbitrarily to the available data centers. While in design pattern distribution the microservices are grouped together on the basis of behavioral design patterns which identifies common communication patterns among objects. The proposed solution was tested using custom built simulation environment and the results showed that the early distribution of microservices according to the design pattern of the application resulted in significant reduction of network calls to the microservices hosted at other network nodes or data centers.