TERA-Scheduler for a Dependency-based Orchestration of Microservices

Due to the increase use of Microservices and the adoption of containers as the best choice for their deployment. Orchestration tools are continuously updating their auto-scaling and scheduling strategies in order to offer better options and capabilities to users. Despite the proposed solutions, there are still many unaddressed issues and optimization is still needed from different perspectives. Therefore, we present TERA-scheduler, with a new scheduling strategy, that we implement using Kubernetes, to reduce both the scheduling duration and the traffic between pods, by considering dependencies between Microservices. Experimentation results show that TERA-Scheduler fasten the response time by 39% compared to Kubernetes (K8s for short) default scheduler which make it a better choice especially for latency-aware systems.