The correct feature is health checks. When a third-party load balancer is placed in front of StorageGRID nodes, it must continuously determine which backend nodes are healthy enough to accept S3 client traffic.
NetApp specifically recommends configuring a health-check mechanism so the load balancer can test each node and remove an unhealthy or unresponsive node from the active pool.
For StorageGRID, NetApp recommends using an HTTP OPTIONS request as a health probe. The load balancer sends the request directly to each Storage Node and expects an HTTP 200 response. If a node repeatedly fails to return the expected response, the third-party load balancer marks that node offline and stops directing new client connections to it. NetApp's guidance gives an example polling interval of once per second and marking a node offline after three failed checks.
Option A, GSLB, distributes traffic between geographically separate sites or data centers rather than determining the health of an individual backend node. Option C, SSL offloading, handles TLS processing.
Option D, session persistence, keeps a client's requests associated with a selected backend and does not detect failed nodes.
Therefore, B. health checks directly satisfies the requirement.
Reference topics: Third-Party Load Balancers # Backend Node Health Checks # HTTP OPTIONS Probe # Node Pool Availability # S3 Client Connectivity