FogEye: Towards Performance Efficient Fog Networking Telemetry.

The concept of fog networking has been introduced to provide Cloud-based services to Internet of Things (IoT) devices. One of the crucial tasks in this paradigm is network telemetry, which plays a vital role in ensuring comprehensive monitoring and management. It helps in coordinating the fog resources between the cloud services and the IoT devices, thus ensuring efficient and effective operations of the entire system. However, the existing network measurement still exhibits some drawbacks, such as a lack of measurement granularities and excessive bandwidth overhead. In the fog network, it remains undefined how to achieve network-wide performance between any links. In this work, We propose FogEye, a network-wide performance telemetry system applied in the fog network. Additionally, we propose a Hierholzer-based algorithm for graph coverage. It effectively reduces the telemetry overhead while ensuring accurate measurement. We conduct experiments in an actual enterprise cloud environment. The result shows that our algorithm can reduce at least 20% of probing packets. FogEye allows operators to query network performance for any link, achieving more cost-efficient traffic monitoring and management under centralized network control.