You are provisioning a new storage container for a production database on a Nutanix cluster. The application requires that data remains accessible and writable even if two nodes fail simultaneously. You have sufficient raw capacity available on the cluster. Which replication factor should you configure for this container to meet the availability requirement without over-provisioning?

AOS uses replication factor (RF) to decide how many copies of each piece of data are written across nodes. RF2 keeps two copies and survives one node failure; RF3 keeps three copies and survives two simultaneous node failures, which is exactly the stated requirement. RF3 needs a cluster configured for fault tolerance 2, which means at least five nodes so that metadata and data can stay available after two losses. RF2 is insufficient because losing two nodes could remove both copies of some extents. Enabling EC-X does not change failure tolerance: erasure coding converts write-cold replicas into data and parity strips that inherit the container's RF, so an RF2 container with EC-X still tolerates one failure (4/1 strip), while an RF3 container with EC-X tolerates two (4/2 strip). EC-X is a capacity saving, not a resilience upgrade. RF4 is not an option in AOS; supported container RFs are RF2 and RF3, with RF1 reserved for specific non-critical use cases. Exam caveat: separate the protection level (RF2 or RF3) from the efficiency feature (EC-X, compression, dedupe) that sits on top of it. Operational check: verify the cluster's redundancy factor and the container's RF in Prism Element, and confirm Data Resiliency status shows two failures tolerated.