Performance evaluation in the migration process from a monolithic application to microservices

Telemonitoring System for Infectious Disease Prediction in Elderly People Based on a Novel Microservice Architecture

“…Microservices represent a relatively new approach to software architectural patterns [16][17][18], as they allow for the development of applications as sets of small services that run independently; in other words, it is not necessary to use the same languages or development platforms [19], [20]. Furthermore, they are interoperable with various communication protocols, since they communicate using lightweight mechanisms such as HTTP [21].…”

“…Likewise, the interactions of each microservice are independent of the physical hosting, and these are either treated individually or grouped into servers or containers [16]. It is therefore important to consider the interaction model that will be implemented in the application, for example for remote procedure calls (RPCs) and event-based interactions [26], the purpose of which is to guarantee that the architecture is robust against failures due to continuous stress at different layers [21], [40].…”

“…It was also necessary to customise the fifth layer of the architecture (infrastructure layer) to support a high level of traffic of clinical data on a daily basis. As a consequence, each microservice has a load balancer (NGINX) to reduce the load for remote calls and in turn to answer users' requests from the application [16]. We therefore use a container orchestrator (Kubernetes) [38] with Docker containers [37].…”

“…Three versions of this project were developed: version Alpha (partial implementation of three microservices, with instances using MariaDB) [5], version Beta v1 (implementation of five microservices with instances using a MariaDB Galera cluster) [14], and version Beta v2 (current implementation of the new microservices architecture, with hybrid instances in MariaDB Galera Cluster and NoSQL), with the aims of achieving the characteristics necessary for the implementation of versatile high-performance applications and identifying the key services for the optimal operation of version Beta v1. The existing components were therefore decomposed and restructured along with their data [16], [18], [52].…”

“…• There are several major constraints that directly affect the performance of microservices, including the network, the organisation of complex services, the consistency of data and transaction management at the database level, and the methods of deployment (bare metal server, virtual machine or container). Consequently, cloud infrastructures play a fundamental role in the operation of microservices and their complexity [16], [19], [21]. It should be noted that the user experience in terms of usability of the operation with the different user services was analysed and validated by both the health personnel and the researchers from the international consortium [9], [14].…”