Nutanix SE Academy NCSE-Core Question # 52 Topic 6 Discussion

To ensure that an RF2 (Replication Factor 2) cluster can survive a complete rack failure without negatively affecting workload performance, it is important to have sufficient nodes distributed across multiple racks to maintain data availability and redundancy.

Current Cluster Configuration:

30 nodes distributed as 10 nodes per rack across three racks.

40TB usable storage per node (total 1.2 PB).

Current utilization: 900TB.

Surviving a Complete Rack Failure:

With RF2, data is replicated to two nodes. To survive a complete rack failure, each rack should have enough nodes to ensure that data can be replicated across the remaining racks.

Adding 3 additional nodes per rack (total 9 nodes) ensures that the cluster can still function with 27 nodes across three racks, maintaining redundancy and performance.

Avoiding Single Points of Failure:

Ensuring that each rack has a balanced number of nodes helps avoid single points of failure and ensures that workloads can continue to operate without performance degradation.

By expanding the cluster to 39 nodes (13 nodes per rack), the administrator ensures that the cluster can handle a complete rack failure while maintaining sufficient storage capacity and performance.

References:

Nutanix Bible, Cluster Resiliency and RF2 Configurations​​.

Nutanix-NDB-User-Guide, Guidelines for Adding Nodes to a Cluster​​.