Do you want the back-end server as a failure pool/server once the original pool fails?
Please refer to the setting guide:
Back-End Fault Tolerance The Traffic Manager has a strong approach to fault tolerance. Pools can be configured in a number of ways to ensure that traffic is managed as efficiently as possible. If a node in a pool fails, this is detected by the Traffic Manager’s health monitors. It can also be detected during load balancing, if the node does not respond to a request. No more requests are sent to that node, but are instead distributed across the other nodes in the pool. When the failed node recovers, it is brought back into service slowly until the Traffic Manager is satisfied that it can be relied upon. A pool can be associated with a failure pool. If every node in the original pool should fail, requests will be diverted to this failure pool. To set a failure pool, click Services > Pools and then click the name of the pool you want to modify. In the "Basic Settings" section, select a failure pool from the dropdown list. A typical failure pool for an HTTP service might consist of a single "sorry server". This can rewrite all requests it receives to request a simple Web page, which displays an "out of service" message.