Arbitrating QoS Factors for Mission Critical Applications at the Edge of the Network
Research project
The digitised society places ever-increasing demands on reliability, security and availability. In response, we are now developing advanced methods to increase security at the edge of the cloud, such as the ability to recover or fully evacuate data.
The rapid development of technology places high demands on the IT infrastructure. Information can no longer be stored on servers far away from users. Mission-critical applications require extreme closeness and accessibility to the system to function. At present, these nodes are more or less unprotected against external influences. Therefore, we are now developing the ability to quickly recreate or completely evacuate data within these edge nodes, while fulfilling high demands on reliability, security and availability.
Truly mission-critical applications at the edge of the network require extreme availability and reliability despite running on nodes that are unprotected from external influences such as weather, accidents or sabotage. To mitigate the effects, support for dynamic protection and recovery is required, such as internal recovery or evacuation of the entire edge node. For such situations, the infrastructure will therefore include significant amounts of processing and storage capacity also at the edge, to meet availability and reliability requirements, such as real-time continuous backup within nodes or real-time replication of processes and data to neighboring nodes.
Switching performance
A positive side effect of the development is that the increased availability of resources provides opportunities for performance improvements, as the system can then trade performance for availability and reliability when resources are available. It may also be an opportunity to increase the future margins for violations of availability and reliability constraints.
Vision and goals
Based on the risk and impact analysis, the challenge now is to dynamically determine the capacity to be used for data processing versus the different protection and recovery tasks.