HYDRA: Decentralized Location-Aware Orchestration of Containerized Applications

Relating Edge Computing and Microservices by means of Architecture Approaches and Features, Orchestration, Choreography, and Offloading: A Systematic Literature Review

“…Several selected studies employ the microservice orchestration by using a container orchestration tool, such as Docker Swarm [124] or Kubernetes [72,75,127]. However, as pointed out by the authors in [16] and [24], the current container orchestration tools do not serve as expected to face the dynamicity of the Edge Computing scenario.…”

“…However, these orchestrators do not scale to the number of nodes that will be required in this new paradigm, nor are they designed to work in the EC [Edge Computing] scenario." - [16] and furthermore in [24]: "Current container orchestrators, such as Kubernetes, are centralized solutions meant for managing local clusters in public and private cloud settings. They have a limited ability to scale, and assume high-performance network connectivity between the nodes, i.e., low latency, high bandwidth and low packet loss."…”

“…• Decentralized Orchestration: It consists of microservices that do not follow a centralized and hierarchical orchestrator. The microservices can work in a full-flat scenario, where every single microservice may orchestrate its peers or organize temporary coalitions to answer a given request [24]. For instance, the authors in [16] propose a decentralized orchestration utilizing different roles: leader, host, controller, and entry.…”

“…However, several authors argue that specific microservice orchestrators are necessary to handle the dynamicity of this environment. That would generate more robust and flexible microservice architectures for Edge Computing [24][25][26].…”

“…A microservice could be offloaded to a different node in the same tier (horizontal) or even on another tier (vertical). The offloading is mainly motivated by performance issues but also could be impelled by geographic constraints, network failures, architecture policy, or topology switches [28,29,24]. Finally, there are three main aspects to consider to build a microservice-based architecture for Edge Computing: the architecture approach and features, the microservice composition, and the computation offloading.…”