Publicação

On the impact of message brokers implementations in the choreography of microservices

Ver documento

Detalhes bibliográficos
Resumo:Communication brokers are essential in modern software development to enable efficient, reliable, and scalable message passing within microservices architectures. However, flawed or delayed communication could be a massive setback that prevents achieving real-time analytics. This paper compares four prominent brokers: Apache Kafka, ActiveMQ Artemis, RabbitMQ, and NATS. Their performance is evaluated in terms of latency, throughput, scalability, and reliability, particularly in the clients implemented in the Java (SpringBoot) and Python languages. Experiments that were conducted in a standardized environment showed that Kafka offers great performance in real-time data processing with its low latency and high reliability. ActiveMQ Artemis provides reliable performance but not without drawbacks as it shows much higher latency. RabbitMQ showed competitive latency but faced some issues in cases of network disruptions. NATS, designed for low-latency and high-throughput scenarios, showed excellent scalability and throughput in all the different scenarios.
Autores principais:Ahmed Gamal Ibrahim
Outros Autores:Lopes, Rui Pedro; Rufino, José; Leitão, Paulo
Assunto:Microservices Message brokers Choreography
Ano:2025
País:Portugal
Tipo de documento:comunicação em conferência
Tipo de acesso:acesso aberto
Instituição associada:Instituto Politécnico de Bragança
Idioma:inglês
Origem:Biblioteca Digital do IPB
Descrição
Resumo:Communication brokers are essential in modern software development to enable efficient, reliable, and scalable message passing within microservices architectures. However, flawed or delayed communication could be a massive setback that prevents achieving real-time analytics. This paper compares four prominent brokers: Apache Kafka, ActiveMQ Artemis, RabbitMQ, and NATS. Their performance is evaluated in terms of latency, throughput, scalability, and reliability, particularly in the clients implemented in the Java (SpringBoot) and Python languages. Experiments that were conducted in a standardized environment showed that Kafka offers great performance in real-time data processing with its low latency and high reliability. ActiveMQ Artemis provides reliable performance but not without drawbacks as it shows much higher latency. RabbitMQ showed competitive latency but faced some issues in cases of network disruptions. NATS, designed for low-latency and high-throughput scenarios, showed excellent scalability and throughput in all the different scenarios.

Atividades financiadas

Carregando projetos financiados...