Disabling NetFlow/CBQoS stops flow data collection, resulting in gaps in graphs during the disabled period (A). However, historical data collected prior to disabling is retained in the database (C), so you can still view past flow information.
Reference:
NTA Administrator Guide, "Enabling/Disabling Flow Collection and Data Retention,"
"Disabling monitoring creates data gaps but does not delete historical flow records."QUESTION

SolarWinds documentation is direct on this point: a red/down interface generally indicates that there is no connection. The first checks should be physical and dependency-based, not custom polling or hardware sensor configuration. The parent node must be reviewed because an interface can appear down or unreachable if the parent device itself is not operating correctly. The cable must also be checked because physical layer failure is a common cause of a down interface. Hardware health sensors are for device component health such as fans, power supplies, or temperature, not primary interface link-state troubleshooting. Device Studio is also irrelevant because the built-in interface status poller already handles interface state.

By default, SolarWinds Network Performance Monitor uses historical data from the last 7 to 30 days to calculate capacity forecasts. This range balances the need for relevant recent data with enough samples to provide trend accuracy.
The documentation states:
"Capacity forecasting in SolarWinds NPM uses the last 7 to 30 days of historical data by default for its projections." References:
NPM Administrator Guide: Capacity Forecast Calculations - Default Data Range