Service Relationship Orchestration: Lessons Learned From Running Large Scale Smart City Platforms on Kubernetes
Merlijn Sebrechtsa b✉, Sander Bornyc, Tim Wautersc, Bruno Volckaertc, Filip De Turckc
- aGhent University-imec, IDLab, Department of Information Technology Technologiepark-Zwijnaarde 126, B-9052 Ghent, Belgium. (e-mail: merlijn.sebrechts@ugent.be)
- bGhent University -imec, IDLab, Department of Information Technology Technologiepark-Zwijnaarde 126, B-9052 Ghent, Belgium
- cGhent University-imec, IDLab, Department of Information Technology Technologiepark-Zwijnaarde 126, B-9052 Ghent, Belgium
- ✉Corresponding author
Papers open inside SciRah from the original publisher or repository; SciRah never hosts them.
Abstract
Smart cities aim to make urban life more enjoyable and sustainable but their highly heterogeneous and distributed context creates unique operational challenges. In such an environment, multiple companies work together with government on applications and data streams spanning several management domains. Deploying these applications, each of which consists of several connected services, and maintaining an overview of application topologies remains difficult. Even though cloud modelling languages have been proposed to solve similar issues, they are not well fit for such a heterogeneous environment because they often require an “all or nothing” approach. Moreover, cloud modelling languages add an additional abstraction layer which rarely supports all features of the underlying platform and make it harder to reuse existing knowledge and tools. This research defines service relationships as the key element to modelling applications as topologies of services. We use this definition to pinpoint what is lacking in the state of the art Kubernetes orchestration tools and provide a blueprint for how relationship support can be added to any orchestrator. We present “orcon”, a proof of concept orchestrator which extends the Kubernetes API to allow managing relationships between services by adding metadata to service definitions. Our evaluation shows this orchestrator enables lifecycle synchronization and configuration change propagation with an overhead of only 0.44 seconds per service.
TL;DR“orcon” is presented, a proof of concept orchestrator that extends the Kubernetes API to allow managing relationships between services by adding metadata to service definitions and shows this orchestrator enables lifecycle synchronization and configuration change propagation with an overhead of only 0.44 seconds per service. (Semantic Scholar)