A Microservice Based Architecture Topology for Machine Learning Deployment

Abstract

Smart solutions that make use of machine learning and data analyses are on the rise. Big Data analysis is attracting more and more developers and researchers, and at least five requirements (Velocity, Volume, Value, Variety, and Veracity) show challenges in deploying such solutions. Across the globe, many Smart City initiatives are using Big Data Analytics as a tool for doing predictive analytics which can be helpful to human well being. This work presents a generic architecture named Machine Learning in Microservices Architecture (MLMA) that provides design patterns to transform a monolithic architecture of machine learning pipelines in microservices with separate roles. We present two case studies deployed to a Smart City initiative, where we discuss how each component of the architecture applied in specific applications that use predictions with machine learning. Among the benefits of this architecture, we argue prediction performance, scalability, code maintenance and reusability makes such transition a natural trend in Big Data and machine learning applications.

Evaluation of Frameworks for MLOps and Microservices

“…Machine Learning in Microservices Architecture (MLMA) is a framework whose objective is to promote some specific design patterns for building microservices with unique responsibilities, that is, segregated from a monolithic architecture [30].…”

“…Specifically, this framework demonstrates two use cases designed for Smart Cities: Tourism Recommendation based on Social Media Photos, which aims to identify the types of environments where the photos were taken and build a preference profile using algorithms in the recommendation system. Predictive Policing, which aims to make spatial predictions related to criminal incidence values for a future time interval [30]. Quantitative runtime data were collected in the Tourism Recommendation based on Social Media Photos use case (Table 1), comparing the execution time in a monolithic architecture and MLMA.…”

“…Table 1. Processing times for recommendation application [30] MLMA is a generic architectural proposal. This architecture enables the implementation of pipelines for ML algorithms, where the framework has the separation of common steps in order to build more adequate services.…”

“…Ref. [30] points out that microservices bring benefits to the maintenance and performance of the framework, where the design makes it possible to work with codes in pipelines of ML algorithms, as well as being an approach that uses microservices. The framework also enjoys other benefits such as, for example, the independence between each of the microservices and, consequently, favouring reuse for a set of different tasks that are present within a specific workflow.…”

“…One of the primary advantages or benefits that has been suggested is the capacity to migrate from a monolithic architecture to an architecture based on microservices, enabling the utilization of ML algorithms in a reusable manner across various tasks within the same workflow. [30] detail the components of MLMA as follows:…”