Codiatech Blog

Workers VPC and mTLS for Private Service Integrations

Cloudflare adds custom CA and client certificates to Workers VPC. We examine private API identities, renewal ownership and failure tests for a controlled company pilot.

A technician checking fiber connectors as an analogy for verifying both network endpoints

What Changed in Workers VPC?

On November 18, 2026, Cloudflare announced custom certificate authority and client certificate support for Workers VPC. Private services can use an internal CA or require mutual TLS. The scope includes HTTPS services and TLS connections such as PostgreSQL accessed through Hyperdrive. This matters to teams connecting cloud applications to systems that use an organisation's certificate infrastructure. Cloudflare announcement

The practical decision is how an integration identifies the server it reaches and how that server recognises the caller. Moving an API connection into a new environment can involve separate owners for network access, certificates and application permissions. A useful pilot brings those responsibilities together in one set of acceptance checks. That makes the resulting connection easier to operate as the environment changes.

Decide What Each Certificate Must Establish

A custom CA concerns the source of trust in the server. A client certificate helps establish the identity of the caller. Before configuration begins, write down the purpose of each component. Verifying an internal API and identifying an integration account are related tasks that need their own acceptance conditions. Assign an owner to each certificate and explain where it is intended to be used.

Cloudflare documents verify_full as checking both the certificate chain and hostname. Workers VPC remains in beta. When a service is updated, certificate references must be preserved in its complete configuration. These details should shape the test environment and rollback procedure. VPC Service TLS documentation

Build an Integration Inventory Before the Pilot

Codiatech recommends a short technical record describing the expected connection. Identify the application, destination service, business operation and environment. Choose a test identity and dataset appropriate to the pilot. Documenting the intended behaviour gives developers and infrastructure teams a shared reference when a result differs from expectations. It also helps the reviewer understand why the integration needs each permission.

  • Record the destination, port and hostname the application will use. Where environments use different names, define the expected certificate match for each one.
  • Inventory the CA chain and validity dates. Assign responsibility for starting renewal and checking the completed change.
  • Describe where the client key is stored, who can access it and how it is deployed. Choose an operational process that keeps key material out of diagnostic logs.
  • Define application success. A working connection should lead to the expected API response or data operation within the pilot's scope.

Test Expected Failures Alongside Successful Calls

A valid certificate and a successful request provide one part of the evidence. Teams should also define when the connection is expected to fail. Run these checks in a separate environment and record the observed result of each scenario. Keep network routing, server identity and application authorisation as distinct troubleshooting questions so a successful change can be explained and repeated.

Check Server Identity

Create separate test cases for an unexpected hostname and an untrusted certificate chain. Define the rejection expected from each case and verify the outcome. Check whether the operator can identify the affected service and certificate from the resulting diagnostic information. When names differ, first compare the DNS entry, configured destination and the names covered by the server certificate.

Check Client Identity and Application Permissions

For a test service requiring client authentication, try a missing or unsuitable client identity. Then use the valid identity for the intended application operation. Review the application's role checks as well. An integration account allowed to read an order can be tested against an administrative operation outside its agreed scope. Keep the test data and expected response documented so another reviewer can reproduce the result.

Rehearse Renewal as an Operational Change

Hands testing a replacement credential with the previous card kept separately

The documentation describes renewal by uploading a replacement certificate and updating the service reference. Service updates replace the complete configuration, so retained TLS fields must be included. A certificate mismatch produces a connection error. These behaviours can be included in acceptance tests. Certificate rotation and configuration rules

Codiatech recommends rehearsing renewal with a controlled stream of requests. Select a low-impact operation, record a successful result before the change and repeat it with the replacement identity. Define the configuration to restore if the change fails. Assess whether returning to an earlier certificate is appropriate for the specific incident before choosing that recovery action.

Monitoring should cover both the expiry date and a real application request. Assigning an owner to the certificate inventory complements a check that exercises the integration. During planned renewal, observe errors, latency and queued work. Use measurements from the company's environment to decide whether the procedure is ready for wider use. An operational review should include who receives the alert and how they identify the affected connection.

An Illustrative Pilot for a Private API

Consider a cloud application that reads a product list from an internal inventory API. A pilot could use a read-only identity and a limited test dataset. Prepare separate acceptance results for valid credentials, an unexpected server name, a missing client certificate and planned renewal. The scenario illustrates a test design and represents no customer deployment or measured Codiatech result.

Before accepting the pilot, have the application owner confirm the data response and the infrastructure owner confirm the certificate change. Keep the test record alongside the deployment configuration. If either owner cannot reproduce the result, resolve the gap before extending the connection to other systems. This gives the team a concrete decision based on its own operating conditions.

Where Codiatech's Services Fit

Codiatech can evaluate data flows and service boundaries through system integration, with operational responsibilities addressed through cloud services and management. The scope depends on the project. Connection validation, application permissions, renewal ownership and incident response should be reviewed together before a production decision. The resulting evidence helps the organisation decide whether to expand the integration.

Sources

Primary source: developers.cloudflare.com