RFT: Scalable and Fault-Tolerant Microservices for the O-RAN Control Plane
Alexandre Huff, Matti A. Hiltunen, Elias P. Duarte · Integrated Network Management · 2021
The Open Radio Access Network (O-RAN) Alliance is opening up traditionally closed RAN elements by defining a new open communication interface (E2) that allows the behavior of a RAN element to be customized and controlled in real time. The RAN Intelligent Controller (RIC for short) is a platform for implementing RAN control functions as microservices called xApps. In this work, we propose and evaluate techniques to enable xApps in the RIC platform to be fault-tolerant while preserving high scalability. The key premise of our work is that traditional replication techniques cannot sustain high throughput and low latency as required by RAN elements. We propose techniques that use state partitioning, partial replication, and fast re-route with role awareness to decrease the overhead. We implemented the fault tolerance techniques as a library, called RFT (RIC Fault Tolerance), that xApp writers can employ to easily make their xApps fault-tolerant. We present performance results which show that RFT meets latency and throughout requirements as the number of replicas increases.