A customer must deploy a new Cisco HyperFlex stretched cluster solution to ensure virtual server availability in case the data center fails. The current network environment provides multiple geographic locations. Which action must be taken to meet these objectives?

A Cisco HyperFlex stretched cluster is designed to maintain availability during a complete site failure. A critical component of this architecture is the witness node. The witness must be placed at a third, independent failure domain (site) with network connectivity to both data sites. Its role is to act as a quorum or tie-breaker in the event of a network partition between the two main data sites. This prevents a "split-brain" scenario where both sites might attempt to become active, ensuring data integrity and consistent cluster operation.

A. The recommended round trip time (RTT) latency for synchronous replication in a HyperFlex stretched cluster is typically 5 ms or lower, not 20 ms.
C. Distance is a secondary consideration derived from latency. The primary constraint is the RTT latency, not a specific physical distance like 300 miles.
D. The data replication factor is a fixed parameter of the stretched cluster design (typically RF=3), not a configurable action to ensure site availability.

1. Cisco, "Cisco HyperFlex Systems Stretched Cluster Guide, Release 4.0," Section: "Stretched Cluster Requirements," Subsection: "Witness VM." The guide explicitly states, "A Stretched Cluster requires a witness VM to act as a tie-breaker... The witness VM must be deployed at a third site."

2. Cisco, "Cisco HyperFlex Systems Stretched Cluster," White Paper, Published: March 2020. Section: "Architecture Overview." This paper details the three-site requirement, with two main sites and a third witness site, to provide quorum and prevent split-brain.

3. Cisco, "Cisco HyperFlex Systems Design Guide," Document ID: 119129, Chapter: "Deployment Models," Section: "Stretched Cluster." It reiterates the necessity of an independent witness for cluster quorum.