While Edge computing is a future reality, the question of how to operate such distributed infrastructures remains. We still miss a resource management system that enables the aggregation/supervision/exposure of all edge resources, and the implementation of new services.
Can OpenStack be extended to satisfy the requirements? How? Our community should address these questions.
Before proposing new pieces of software, it is essential to conduct a gap analysis of the current code base. This talk presents the study carried within the FEMDC SiG.
First, we introduce capabilities admins/users may expect - from the simplest, start a VM at a specific location; to the most advanced, interoperability across cloud stacks. Second, we discuss how we evaluated the different deployment scenarios that might be envisioned with the current OpenStack code base. We conclude by presenting the premise of a possible architecture leveraging OpenStack with the goal of initiating the debate with the community.
- An overview of edge infrastructure deployment scenarios
- An overview of the administrators/end-users' capabilities an edge resource management system should fulfill (e.g. start a VM in a certain location, cross-site networks, migrate workloads across sites)
- An analysis of the pros/cons of the different OpenStack deployment scenario (cells, regions, federation...)
- The premises of possible software architectures of a resource management system for edge computing infrastructures.