A backend VM group terminates long-lived TCP connections and you need to limit the number of simultaneous connections routed to each VM. Which backend service balancingMode should you choose?

The CONNECTION balancingMode is designed for scenarios involving long-lived TCP connections, as it directly limits the number of simultaneous connections sent to each backend instance. This ensures that no single VM is overwhelmed, maintaining stability and performance. UTILIZATION balances based on CPU or other resource usage, which is less effective for long-lived connections where resource consumption might fluctuate independently of connection count. RATE balances based on requests per second, suitable for HTTP(S) load balancing, not raw TCP connections. NONE is not a valid balancing mode option.