These days I have written a report on the relationship between IAM (Identity and Access Management) and SOA (Service oriented Architecture/Applications). One major aspect of this relationship is around end-to-end-security, e.g. securing the interaction of a user with an application (and the application which implements a business process) up to the backend systems like databases.
That is inevitable because using a service in the context of an user identity or an user role is the only way for consistent, externalized security instead of coded security where some return of a service is filtered by the application depending on the user’s role. Coded security is contradictory to compliance, obviously. It’s expensive in terms of coding and auditing. Thus, it doesn’t make sense.
On the other the most common approaches for web service security are constructed the same way as web access management solutions: Building a layer in front of the services which uses policies to decide how services are used. That includes some part of authorization and sometimes authentication. The problem is: Using such an approach means that there is definitely no end-to-end-security. From my point of view, there is no alternative to federation to transport claims down to the service level. That is the only approach for real end-to-end-security and thus for applications which are architected to fulfill the increasing compliance requirements.