What you are describing is all automatic in Teradata. If an AMP is down, and fallback is in place, then the query automatically reads the fallback copy and reads all current data. If updates are performed, they are applied to the fallback copy automatically and queries raed those updates as tehy would in the primary copy. The updates are also saved to be reapplied to the primary when it is available. When the AMP comes back up, the data for the primary is recovered from fallback and from any saved updates until the primary is again equivalent to the fallback copy - all automatically. While this recovery is going on, any queries automatically continue to read the fallback copy. When the AMP is caught up with its recovery, a short system restart is performed to return the failed amp to the configuration and queries then automatically go back to reading the primary copy.
What you are describing is all automatic in Teradata. If an AMP is down, and fallback is in place, then the query automatically reads the fallback copy and reads all current data. If updates are performed, they are applied to the fallback copy automatically and queries raed those updates as tehy would in the primary copy. The updates are also saved to be reapplied to the primary when it is available. When the AMP comes back up, the data for the primary is recovered from fallback and from any saved updates until the primary is again equivalent to the fallback copy - all automatically. While this recovery is going on, any queries automatically continue to read the fallback copy. When the AMP is caught up with its recovery, a short system restart is performed to return the failed amp to the configuration and queries then automatically go back to reading the primary copy.