SMO

The SMO is the management umbrella over an O-RAN deployment — the layer that operates, configures, and orchestrates everything else. It hosts the non-RT RIC (and its rApps), and it reaches the rest of the network through a couple of distinct interfaces: O1 for FCAPS-style management of the RAN nodes (fault, configuration, performance, the day-to-day operational data), and O2 toward the underlying cloud platform (the O-Cloud) that the RAN functions actually run on.

Drawing that line between O1 and O2 is worth getting straight. O1 manages the network functions — the O-CU, O-DU, near-RT RIC themselves. O2 manages the infrastructure they're deployed on — provisioning compute, lifecycle of the virtualised resources. The SMO ties both together so an operator can, in principle, run a multi-vendor RAN from one management plane. In practice the SMO often overlaps with or sits beside an operator's existing OSS and orchestration stack, which is part of why its exact boundaries vary between deployments.