1Z0-1072-25 Question 138
Single answerYou manage a high-throughput HPC workload running on a private subnet behind an Oracle Cloud Infrastructure (OCI) Load Balancer. The HPC application tracks detailed connection metrics and needs the original client IP address for analytics. The application uses a custom protocol (not HTTP) over TCP. Which configuration choice ensures that the HPC nodes receive the true client IP address?
- A
Set up an HTTPS listener and rely on X-Forwarded-For headers to transmit the client IP.
- B
Configure a TCP listener in pass-through mode and select the option to preserve the client IP.
- C
Use a public load balancer with a default Round Robin policy and rely on the system logs for client IP addresses.
- D
Place the HPC nodes in a public subnet, remove the load balancer, and directly expose node IP addresses to clients.
Show answer and explanation
Correct answer: B
Explanation
In OCI, preserving the client� source IP address for non-HTTP workloads requires using a TCP listener with pass-through mode. HTTP and HTTPS listeners typically terminate the connection at Layer 7 and rely on adding X-Forwarded-For headers to forward the client� IP. Since this HPC workload is not using HTTP, the recommended solution is a TCP load balancer with 'preserve client IP' enabled. Refer to the 'OCI Documentation on Load Balancer Listener Types' for detailed configuration instructions.
- A. Incorrect.
Incorrect. X-Forwarded-For headers apply to HTTP or HTTPS traffic, not custom TCP protocols. The HPC application would not see original client IP addresses via HTTP headers.
- B. Correct.
Correct. A TCP (Layer 4) load balancer with pass-through mode configured to preserve the client IP is the supported method in OCI for transmitting the true client IP address when not using HTTP/HTTPS.
- C. Incorrect.
Incorrect. Merely choosing a Round Robin policy does not guarantee the client IP is preserved. Also, system logs do not circumvent NAT if the load balancer terminates and reinitiates traffic at Layer 7.
- D. Incorrect.
Incorrect. Exposing the HPC nodes in a public subnet bypasses the load balancer entirely. Although the HPC nodes would see client IPs, it sacrifices the benefits of load balancing (like health checks, scalability, and centralized TLS if needed).