Discover Smile's R&D in cloud orchestration, machine learning, and 5/6G slicing. Explore our digital services from design to development and maintenance.
Resources scheduling
In terms of digital infrastructure, the past 10 years were the “Cloud” years : many if not most customers and editors moved from private, on premise, infrastructures, to public, Cloud, ones. This migration effort was meant to enable an ever-growing digital capacity as well as optimizing DSI costs.
As the public Cloud usage grew, so did software distribution, with :
- “As a service” approaches, where applications can be integrated together through APIs calls (enabling new architectural concepts known as serverless).
- Apparition of new methodology along with an ecosystem of tools known as DevOps (which came with their own virtualization and packaging technologies, known as containers).
- An expanding mobile usages (requiring fine scalability server side to support availability, notably of free services)
For these reasons, Cloud infrastructures evolved towards new concepts known as Edge and Fog.
While Cloud represents a centralized infrastructure, that can be hosted at a region, country or continent level, Edge represents a decentralized infrastructure, that is hosted closer to users, at a city, street or building level. Edge infrastructure supports the Cloud infrastructure by providing either resiliency, through services instances and data copies, or throughput, through dedicated networking devices and services, such as antennas or accelerators.
Cloud and Edge collaborate together to provide a satisfying experience to end users, or support services provided by IoT infrastructures. This collaboration is known as the Cloud/Edge continuum : although the Cloud and Edge infrastructures can operate at different places, time scales or from various administration teams, there always is an ongoing continuation in its operation.
Scheduling the Cloud/Edge continuum, that is automating the collaboration of two or more infrastructures, is an actively researched topic, in which Smile R&D proposes different solutions aiming at either simplifying Cloud/Edge optimization or understand better the current state of execution. To quote a few : continuous optimization of resource placement based over time series analysis, high level domain indicators derived from system metrics, applications lifecycle modeling…
You can find most of our open source research results within the SMILE GitHub group : https://github.com/Smile-SA
If you want more information, feel free to drop us a mail at rnd@smile.fr