A name identifier, represented by the
<NameIdentifier> element in SAML1 and the
<NameID> element in SAML2, is generally used to identify a direct way to name the subject of a SAML assertion. Name identifiers can be anything; : an email address or , a Kerberos principal name are common, everyday examples of such information. SAML2 , a certificate subject, an employee ID, a username, or literally anything else. SAML 2.0 also defines more specialized identifier types with particular properties that were presumed useful in federated applications.
Unfortunately this generality makes interoperability much more complex than one would prefer.
Strictly speaking, SAML assertions don't have to contain an a name identifier. The subject may be implicitly identified as the bearer of the token or anybody able to demonstrate possession of a key. In SSO use cases, one reason for including an identifier is to enable the relying party to refer to the subject later, such as in a query, or a logout request. So-called "transient" identifiers that are generated uniquely for each assertion are often used to support those use cases and are a common pattern in Shibboleth deployments.
Every name identifier is associated with a format. Formats label the identifier at runtime to help applications process them appropriately. They're conceptually similar to an Attribute Name and in fact one conventional way to express a SAML Attribute as a name identifier is to encode its Name as a Format (assuming the Attribute Name is a URI).
Though this is a retroactive view of the design, Name identifiers can be described by the following characteristics:
A special type of globally unique identifier is a scoped attribute, which has the form
userid@scope. In practice, the scope value is a DNS domain, which ensures global uniqueness.
Here are some examples (not all of these are actually encoded as SAML name identifiers, some are defined solely as Attributes):
|Identifier / Attribute||Persistent||Revocable||Reassignable||Opaque||Targeted||Portable||Global||Qualifier|
|SAML2 Transient NameID||No||N/A||N/A||Yes||N/A||N/A||Yes||N/A|
|SAML2 Persistent NameID||Yes||Yes||No||Yes||Yes||Yes||No||Issuer ID|
|SAML2 Subject ID Attribute||Yes||Yes||No||Yes||No||Yes||Yes||Scoped|
|SAML2 Pairwise ID Attribute||Yes||Yes||No||Yes||Yes||Yes||Yes||Scoped|
|Social Security Number||Yes||No||N/A||No||No||Yes||No||US Citizens|
|OIDC public ||Yes||Yes||No||N/A||No||No||No||Issuer ID|
|OIDC pairwise ||Yes||Yes||No||N/A||Yes||No||No||Issuer ID|
- The SAML2 Persistent name identifier and the eduPersonTargetedID attribute are functionally equivalent. Indeed, the value of the latter is precisely a SAML2 Persistent
- The SAML2 Persistent name identifier (and hence eduPersonTargetedID) are portable in the sense that any issuer can assert a known SAML2 Persistent
<NameID>element. For example, a SAML2 Persistent
<NameID>can transit a SAML IdP Proxy as-is, without modification. The same applies to the newer SAML2 Subject ID and Pairwise ID Attributes.
- The SAML2 Persistent name identifier and the OIDC pairwise
subclaim differ with respect to the portability characteristic only. In particular, the
subclaim can not transit a gateway since the
issclaim is required for global uniqueness.
- A Phone Number is not universally portable but within the US, Phone Number is indeed a portable identifier. In fact, it is one of the few portable identifiers with no qualifier.
Shibboleth deployments traditionally have focused on the use of Attributes to describe subjects, and default to the use of transient name identifiers (or omitting them). Commercial SAML deployments rarely less commonly make use of Attributes and tend to use loosely or improperly specified name identifiers.
The properties above used to describe name identifiers also apply to attributes when those attributes are themselves unique identifiers for a subject. Of course, many attributes are not identifiers at all, merely data of various kinds.Note, however, that using this mechanism does not magically convey the reversibility property on an attribute. Whichever attribute is chosen to be encoded as the name identifier must already have this property. The encoding process does, however, add the scoped property and is not something that a deployer must explicitly configure.