Brownfield Measurement: A Practical Grey Failure Identification and Localization Method in Incremental Deployment Network

In recent years, the In-band Network Telemetry (INT) method with fine measurement ability has been gradually deployed in various datacenter networks. However, in real-world production practice, it is difficult to replace all devices simultaneously, and an essential requirement is to support the brownfield deployment, which means replacing the devices gradually. Considering the above situation, due to the "black switch", a nonprogrammable switch, it is difficult for existing measurement methods to independently identify and locate grey failure in time. In this paper, we propose "Brownfield Measurement" (BMM), a practical identification and localization method of grey failures in the incremental deployment environment. This method aims to integrate the INT and network tomography (NT) technology with a redesigned collaboration measurement mechanism and penetrate the advantages of INT into the "black region". At the same time, considering that the original INT methods based on the central control model are prone to controller bottlenecks due to large overhead, BMM is also designed as a method based on edge device measurement, and some tasks that originally needed to be run by the controller are delegated to the network element devices located at the edge. This not only reduces the burden on the controller, but also effectively reduces additional overheads. Finally, our experiments prove that the proposed method has good identification performance with key indicators, such as latency and packet loss rate in the brownfield deployment environment, to realize quick failure localization with a great precision rate at the edge switches.