You have decided to connect your service to SURFconext. Instead of using a username and password, your service will be dealing with user attributes with SAML or claims if you use OpenID Connect. But what attributes or claims should you request?
As a guideline, use:
- As few as possible for your service to work properly.
- As privacy preserving as possible for your service to work properly.
SURFconext, in line with the European General Data Protection Regulation (GDPR), assumes a minimum disclosure principle. In short, this means a service should only process (personal) information that is strictly necessary for providing the service. On this page you will learn more about:
In this webinar (Dutch) we explain the important identifiers available in SURFconext
Identifiers: the best option
From a privacy perspective, the best possible way to make your service work within SURFconext is to use a pseudonymised identifier generated by SURFconext. In SAML, this value can be found in the NameID element; in OpenID Connect, it will be in the sub claim. The use of this identifier is both privacy aware and robust.
When using this identifier, there are two options: transient and persistent.
|SAML Attributes||OpenID Connect Claims||Explanation|
|NameID element (transient)|
sub claim (transient)
|Service Providers who have no need to identify one user from another should use this element. The transient NameID/sub claim is an opaque identifier generated by SURFconext. It is impossible to see who is behind it. It also changes every time the user re-authenticates. As a Service Provider, you are therefore unable to see whether it is the same user logging in, or someone else.|
|NameID element (persistent)||sub claim (persistent)|
If you do need to identify returning users, the persistent NameID/sub claim is available; a persistent NameID/sub claim is like the transient NameID/sub claim explained above, except that returning users will have the same value each time they return. This attribute is privacy aware and is, of all attributes available, least likely to change for a given user thus a certain and consistent identifier.
Identifiers: the second best option
If your service needs more information, for example if your service needs to be able to recognise different type of users on their first login, we can provide additional identifiers. In order of order of privacy preserving, the following applies:
|SAML Attribute||OpenID Connect Claim||Explanation|
|schac_personal_unique_codes||This is a number that is often used within internal institution systems and very well suited for uniquely identifying users without affecting their privacy too much. However, currently only a few institutions provide this attribute. If you wish to use this attribute, please inform the SURFconext Team; they will ask the institution to provide this attribute.|
A uid is unique for an institution, but it is not unique within SURFconext. The same uid could be used for different persons in different institutions. Be aware that uid is reusable. This means that when 'John' leaves the university, a second John could be provided with the same uid. This attribute can be useful in the scenario that your services uses the value of the uid from the institution for identification. This attribute gets globally unique when it is combined with the schacHomeOrganization (see below)
|The principal name can be used to uniquely identify a user. Downside is that in most cases, the value gives away a lot of information about the user. Bear in mind that the principal name is reusable.|
|Mail||email||The email attribute could also be used to identify a user. However, we strongly advise against it: an email address can change over time (e.g. after marriage or when someone changes surname). Also, email is reusable. If email@example.com leaves the university, a second John could be provided with the same email address. Furthermore, the email attribute is multi valued, which means a user can have more than one email address.|
Next to identifiers, there are other types of attributes available to discern between different types of users without necessarily knowing the exact identity of the user.
Mapping attributes to existing users
If you are connecting an existing service you probably already have a user database. You need attributes which you can map to the information you already have about users. Common options are:
Minimize authentication attributes
Please consider using as few and privacy friendly attributes for the authentication flow and present the user an option after he/she has authenticated, so has entered your service, to supply you with more profile information, for instance in a user account management panel. We assume, when doing that, you also keep in mind with all laws and regulations, for instance regarding consent.
For all available attributes and technical details read on on our attributes page in our backgrounds section. The translation from attributes to claims can be found there as well. Remember, the more attributes you request, less obvious it becomes institutions are willing to connect to your service. Choose wisely.