RAINBOW partner TUW will present their paper “Polaris Scheduler: SLO- and Topology-aware Microservices Scheduling at the Edge” [1] at the 15th IEEE/ACM International Conference on Utility and Cloud Computing (UCC2022), which will take place on December 6-9, 2022 in Vancouver, Washington, USA. Polaris Scheduler, is a fork of the RAINBOW Fog-aware Kubernetes scheduler in the Centaurus project. Due to the continuous exchange between RAINBOW and Centaurus both schedulers are currently almost identical.
The main goal of Polaris Scheduler and RAINBOW scheduler is to place microservices in a dynamic Edge/Fog environment to fulfill not only the resource requirements, but also the network Quality of Service (QoS) Service Level Objectives (SLOs). To this end, users leverage the RAINBOW Service Graph abstraction to model the dependencies between their applications’ microservices and optionally annotate the dependencies with network QoS requirements, e.g., minimum bandwidth or maximum allowed latency, as shown in the diagram above. The scheduler relies on a Cluster Topology Graph to provide information about the current network connections between the cluster’s nodes. This graph needs to be continuously updated by monitoring solutions active in the cluster. The scheduler uses both, the Service Graph and the Cluster Topology Graph, in conjunction to compute a placement that will allow the application to fulfill its network SLOs.
The paper uses a Traffic Analysis & Hazard Detection scenario with stringent network QoS requirements to motivate the need for network SLO adherence in scheduling and for evaluating the effectiveness of the scheduler. This scenario is heavily inspired by RAINBOW Use Case 2: “Digital Transformation of Urban Mobility”.
Details on the scheduler’s design and the employed algorithms, as well as the full evaluation, will be presented at the UCC2022 conference and will be available in the full paper shortly afterwards.
References
[1] T. Pusztai, S. Nastic, A. Morichetta, V. Casamayor Pujol, P. Raith, S. Dustdar, D. Vij, Y. Xiong and Z. Zhang, “Polaris Scheduler: SLO- and Topology-aware Microservices Scheduling at the Edge,” in 2022 IEEE/ACM 15th International Conference on Utility and Cloud Computing (UCC), 2022.
