Name of the participant: Florian Brandherm
Description of the IT research project: Cloud computing means concentrating the data centre needs of a large number of customers in a few, very large data centres. While sharing does exploit economies of scale, cloud data centres are often far away from their end users. This distance creates a significant communication delay and prevents potential services that rely on low latency from benefiting from the economics and flexibility of the cloud. In addition, the centralised architecture creates massive costs for network operators as more and more traffic needs to be routed between these data centres and end users.
To mitigate these shortcomings of centralised data centres, the concept of edge computing was introduced. The idea is to complement today’s centralised cloud data centres with a multitude of geographically dispersed micro data centres. Since these are built very close to the end devices (e.g. mobile base stations), they can offer significantly lower latency as well as cheap bandwidth. This will enable fundamentally new cloud-like services, such as virtual-reality-as-a-service, services for autonomous driving or services for Industry 4.0.
However, it has not yet been finally clarified how such a highly distributed, yet highly critical infrastructure is ideally managed. Unfortunately, the wealth of experience in orchestrating large centralised data centres can only be partially reused in this scenario, as edge computing brings new challenges. For example, many endpoints are expected to be mobile, requiring dynamic migration of services to keep up with their latency or bandwidth requirements.The goal of the project is to increase knowledge about the placement and migration of Critical Services using machine learning.
Software Campus partners: TU Darmstadt, Huawei
Implementation period: 01.01.2021 – 31.12.2022