That is one of the main use cases for GTM.
You would basically configure a pool in each data center, and change the load balancing type to Global Availability. Then set a higher priority to the "HO" data center than the "DR" data center.
With that configured everything will be sent to "HO" until it becomes unavailable.
HERE is a post asking something similar to what you are asking. Read through it as there is some good info!
Hope that helps! If it does please up-vote and select this answer, it'd be greatly appreciated!