Deployment Considerations for the R&S Category

Service Providers

It is important that the deployment of all InCommon services facilitate initial on-boarding processes to avoid operational and technical impediments to adoption, as described in Recommended Practices for InCommon Participants.

More specifically, R&S services generally have a broad user community, often including people who do not have a close relationship with the Service Provider, or whose IdP operators do not have a close relationship with the Service Provider. For this reason, R&S Service Providers are encouraged to consider the following guidelines:

Although R&S is specifically designed to facilitate attribute release, errors are expected and therefore service providers are strongly encouraged to support Federated Error Handling. A centralized Error Handling Service is provided for this purpose.

Identity Providers

An IdP supports R&S if the IdP releases some subset of the R&S attribute bundle to R&S SPs without administrative involvement, either automatically or subject to user consent. A list of IdPs that are known to support R&S is maintained elsewhere in this wiki.

To support R&S, an IdP has at least three options (in increasing order of deployment difficulty):

  1. Release a fixed subset of the R&S bundle (or the R&S bundle itself) for all SPs
  2. Release a fixed subset of the R&S bundle (or the R&S bundle itself) for all R&S SPs
  3. Release a precise subset of the R&S bundle for each R&S SP (on an SP-by-SP basis)

The Shibboleth IdP software supports either of the first two options out-of-the-box. The latter option requires a special plugin at the Shibboleth IdP. No other IdP software is known to support entity attributes at this time.

Sites are strongly encouraged to configure their IdPs to support R&S, either by releasing R&S attributes directly to R&S SPs or releasing directory information to all SPs.

The R&S category is the first of many such categories. Soon there will be multiple categories, for both SPs and IdPs, such that each category has its own entity attribute value. To support a given category, an additional software configuration similar to the R&S configuration is required.

The use of entity attributes (as opposed to entity IDs) has a significantly reduced administrative burden at the IdP. As the number of categories increases, however, the number of configurations increases as well. It is natural to ask if there is an even higher level of abstraction that further simplifies the administration of attributes? Yes, an IdP can release directory information to all SPs, not just R&S SPs. Such a configuration can simultaneously satisfy the attribute requirements of multiple categories.