Your question is High Availability and Data Durability. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
Tell me how you would ensure high availability and data durability in a distributed system.
Discuss the design of an end-to-end ML platform, including online inference, feature and training data storage, replication, failover, recovery, and monitoring. Explain the tradeoffs among consistency, availability, durability, recovery point objective, and recovery time objective.