Building an IP allowlist for a partner API integration

Coordinate a static proxy IP allowlist with a partner API, verify the observed address and prepare for address renewal or replacement.

A truck approaches a raised barrier and staffed gatehouse outside an office complex.

Match the source your partner actually sees

An API partner normally applies its access rule to the incoming connection it receives. That may be different from the address assigned to a developer's laptop or the address of the gateway the developer connects to. Ask the partner to identify the expected egress family and its process for making changes before requesting an allocation.

Coordinate the first test

Prepare a harmless endpoint or read-only request agreed by both sides. Give the partner the approved egress addresses through its normal change process. Record when the rule takes effect and which environment it applies to. Confirm receipt with a test identifier rather than sharing sensitive request bodies. If staging succeeds but production does not, check the scope of the partner's rule before changing your whole application.

For a long-running project, put a reminder on the calendar before both the proxy allocation and the partner rule expire. Those dates can differ even when the two services were first configured on the same day. MDN guide to proxy servers and tunneling.

Plan for renewal and replacement

Agree on who owns the access list, who receives address-change notices and how much lead time the partner needs. Use the same approval process to remove obsolete entries when an allocation ends. When asking XtremeProxy about static egress, include the required address family, term and notice period. A modestly priced IP can carry substantial operational value when the partner's change process is slow.

A partner approves the address seen at its boundary

A provider assigns a fixed endpoint to a test integration. The worker's own local address differs from the address the partner sees after outbound translation. Ask the partner to confirm the source for one approved request using a synthetic correlation identifier. Add only the jointly agreed address or prefix to the intended staging rule. Record the environment and expiry date before promoting the integration through the partner's normal approval process.

Rehearse a controlled replacement

A partner can require advance notice for a rule change and a separate approval for production. Document the existing source, proposed replacement, agreed overlap and time zone. Use restricted staging rules for the rehearsal and retire the old entry on schedule. Never broaden a partner rule merely to make the acceptance check pass.

Make an explicit owner responsible for removing retired addresses. An integration can continue working for months with an obsolete address left on the partner’s list. Tie each approved entry to an assignment record and review date so both teams can remove it promptly when the address is released.

Verify both introduction and removal

A partner temporarily authorizes two approved sources during a controlled change. One successful request from the replacement verifies introduction. After the agreed overlap, the partner confirms removal of the previous source under its normal procedure. Those are separate completion conditions. Keep the date and policy environment so an old staging rule cannot be mistaken for a live authorization.

If the partner maintains separate allowlists for sandbox and production, a sandbox acceptance ticket is not evidence that a production change has been approved. Give each entry an environment, access owner, approval reference and intended effective date. Before the production cutover, have the partner confirm the expected source and service in its normal change process. Conflating environments can lead both teams to debug an otherwise sound network configuration while the relevant approval is still in a different queue.

Ask whether the partner evaluates the source before or after its own gateway. An application log may show an internal inbound proxy, while its access-control decision was made by a separate component at the perimeter. Have the partner identify the actual evaluation point and share a redacted decision record for your synthetic staging request. The result is stronger than comparing two public-address displays that may describe different legs. Treat any forwarded source fields according to the partner's documented trust and configuration rules.

Keep the partner's own observation with the assignment and renewal record. That gives the operational team a test it can repeat safely when the source changes. Partner inventory feed allowlists walks through a partner feed with a named source. When a static ISP proxy is worth paying for explains when address continuity justifies the additional cost and planning of a fixed service. Technical references include RFC 8200, Internet Protocol version 6. Compare XtremeProxy network types.

Frequently asked questions

Can a partner use the gateway hostname directly as an address rule?

That depends on its supported control and how the hostname resolves over time. Confirm the value the partner actually evaluates at its boundary and the change process for replacement assignments.

Related articles