A Resource Awareness Mechanism for Power Industry Computing Force Network

• Published in: 2023 15th International Conference on Communication Software and Networks (ICCSN) pp. 66 - 73
• Main Authors: Zhang, Peiming, Li, Bo, Liu, Yuanjie
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21.07.2023
• Subjects: computing force network; Costs; Force; in-band network telemetry; Jitter; Metadata; Packet loss; Processor scheduling; Real-time systems; resource awareness mechanism; vector packet processing
• ISSN: 2472-8489
• DOI: 10.1109/ICCSN57992.2023.10297334

In the face of various computing business requirements of the Electric Internet of Things (EIoT), it is necessary to comprehensively consider the distribution and scheduling of computing and network resources in the existing electric power data network in the cloud-network-edge fusion scenario, which is a typical computing force network (CFN). CFN is a new information infrastructure that provides integrated services based on the deep integration of computing power and network. It can achieve flexible on-demand scheduling of business based on the needs of computing business and real-time computing and network resource state. Real-time perception of the state of computing and network resources is a key technology in CFN. However, there is currently no unified resource awareness solution, especially in the EIoT scenario where computing and network resources are limited. In this work, we designed a resource awareness mechanism for CFN in EIoT by improving INT (In-band Network Telemetry) technology, and implemented a resource awareness system using VPP (Vector Packet Processing) platform. This resource awareness mechanism is designed on the basis of a new INT mode that we proposed, namely INT-HX (Hybrid Data Export) mode. The INT-HX mode exports INT Metadata through a hybrid approach, which supports hop-by-hop measurement of delay, jitter, and packet loss rate while reducing the export cost of INT metadata. In addition, this mechanism can complete the notification of computing resource state information through the forwarding of INT packets without generating additional packets. We have verified the availability of the system and the effectiveness of this mechanism through experiments. The experimental results show that the system can effectively perceive the state of network and computing resources. This mechanism supports the carrying and forwarding of computing resource state information, as well as the measurement of hop-by-hop packet loss rate, and reduces the export cost of INT Metadata.