Service Consumer A sends a request message with an authentication token to Service A, but before the message reaches Service A, it is intercepted by Service Agent A (1). Service Agent A validates the security credentials and also validates whether the message is compliant with Security Policy A .If either validation fails, Service Agent A rejects the request message and writes an error log to Database A (2A). If both validations succeed, the request message is sent to Service A (2B). Service A retrieves additional data from a legacy system (3) and then submits a request message to Service B Before arriving at Service B, the request message is intercepted by Service Agent B (4) which validates its compliance with Security Policy SIB then Service Agent C (5) which validates its compliance with Security Policy B .If either of these validations fails, an error message is sent back to Service A .that then forwards it to Service Agent A so that it the error can be logged in Database A (2A). If both validations succeed, the request message is sent to Service B (6). Service B subsequently stores the data from the message in Database B (7). Service A and Service Agent A reside in Service Inventory A .Service B and Service Agents B and C reside in Service Inventory B .Security Policy SIB is used by all services that reside in Service Inventory B .Service B can also be invoked by other service consumers from Service Inventory B .Request messages sent by these service consumers must also be compliant with Security Policies SIB and B .Access to the legacy system in Service Inventory A is currently only possible via Service A, which means messages must be validated for compliance with Security Policy A .A new requirement has emerged to allow services from Service Inventory B to access the legacy system via a new perimeter service that will be dedicated to processing request messages from services residing in Service Inventory B .Because the legacy system has no security features, all security processing will need to be carried out by the perimeter service. However, there are parts of Security Policy A that are specific to Service A and do not apply to the legacy system or the perimeter service. Furthermore, response messages sent by the perimeter service to services from Service Inventory B will still need to be validated for compliance to Security Policy B and Security Policy SIB .How can the Policy Centralization pattern be correctly applied without compromising the policy compliance requirements of services in both service inventories?