4.2 Management Cluster Operation Restrictions
The vSAN cluster with two data nodes and one witness has the following restrictions during regular and extraordinary operations:
• When a data node fails, the workloads are still operative, but no attempt to rebuild the lost data can be made until the failure is resolved or the failed node is replaced. The drawback of a small two-node configuration is the reduced reliability after the first failure.
• When a data node is placed in maintenance mode, the same behavior applies until the node is back online.
• During regular operation, storage I/O is split equally across the two data nodes. During an outage or maintenance, 100 percent of I/O is sustained by one data node and the storage performance must be sized to sustain that particular load.
Note If any of these constraints are not acceptable for the SLA in place with the tenants or with the end users, a larger deployment with three data nodes must be considered as the reference configuration instead.