We need to decide whether there should be a separate model just for metadata semantics between nodes and monitors, or whether the yang model should cover all of data plane telemetry including normal management and configuration.
In openconfig, they strongly advocate a separation of config and operational state. If the scope remains narrow, then we won't have any config state, only operational state, so it deviates from openconfig style to some extent.
We need to decide whether there should be a separate model just for metadata semantics between nodes and monitors, or whether the yang model should cover all of data plane telemetry including normal management and configuration.
In openconfig, they strongly advocate a separation of config and operational state. If the scope remains narrow, then we won't have any config state, only operational state, so it deviates from openconfig style to some extent.