Several partner agencies need private access to a municipal API without building a full VPC peering mesh and extra NAT paths. Which approach can reduce cost and complexity?
Select an answer to reveal the explanation.
Short Explanation
Don't build a spaghetti of driveways to every partner garage. PrivateLink gives partners a private front door without a full peering mesh and NAT maze.
Full Explanation
When several partner agencies need private access to a municipal API, a full VPC peering mesh scales poorly: each new partner adds routes, security group references, and often extra NAT paths for hairpinned traffic. AWS PrivateLink exposes the API as a Network Load Balancer-backed endpoint service so consumers create interface endpoints in their VPCs and reach the service privately over the AWS network. That reduces peering-link sprawl and NAT hairpins while keeping traffic off the public internet—aligned with inter-agency data-sharing constraints for permitting, benefits, or records APIs.
Peering every partner VPC to every other VPC and adding NAT Gateways in each spoke fails by concept: mesh complexity and NAT hours grow combinatorially. Exposing the API on the open internet with a wide security group and no authentication abandons private connectivity and security goals. Shipping nightly API responses on Snowball Edge to each partner loading dock confuses bulk offline migration with interactive API access.
Exam caveat: PrivateLink consumers pay for interface endpoint hours and data processing; it is still usually cheaper operationally than a brittle mesh, but it is not "free private networking." Provider and consumer accounts must manage endpoint service acceptance and least-privilege endpoint policies.
Operational check: publish the municipal API as an endpoint service, have one partner pilot an interface endpoint, verify private DNS and connectivity without peering, compare NAT and peering operational tickets before versus after, and review interface endpoint charges against the avoided mesh cost in Cost Explorer.