Federation and Single Sign-On

How federation and SSO reduce account sprawl while preserving clear trust, authorization, and lifecycle controls.

Single sign-on and federation are often mentioned together, but they are not identical. Single sign-on is the user experience: sign in once and reach several applications. Federation is the trust relationship: one identity provider authenticates a principal so another service can rely on that result.

The practical value is not merely fewer passwords. The value is central control over identity lifecycle, authentication strength, account status, groups, claims, and sign-in evidence. Without that governance, SSO can become convenient account sprawl with a nicer login screen.

The federation pattern

    sequenceDiagram
	  participant U as User
	  participant SP as Service Provider
	  participant IdP as Identity Provider
	  U->>SP: Request application
	  SP->>IdP: Redirect or request authentication
	  IdP->>U: Authenticate user
	  IdP->>SP: Send signed assertion or token
	  SP->>SP: Map claims to session and permissions
	  SP->>U: Allow application access

The identity provider proves the user. The service provider still needs to map identity attributes, groups, claims, roles, or scopes into application authorization.

SSO and federation chooser

Requirement Strong answer
Employees need one login for many enterprise apps SSO through a central identity provider
A cloud provider should trust the corporate directory Federation between the cloud account and the enterprise IdP
A SaaS app needs user attributes such as department and group Federated claims mapped to application roles
A contractor should lose all app access when engagement ends Central lifecycle deprovisioning through the identity source
An app needs delegated API access on behalf of a user OAuth-style delegated authorization, often with OIDC for sign-in

The strongest design usually avoids local user copies in every target application. Local accounts become stale, difficult to audit, and easy to miss during offboarding.

Trust must be scoped

Federation is a trust relationship, not a blank check. A well-scoped federation design controls:

  • which identity provider is trusted
  • which application, audience, or relying party the assertion is intended for
  • which claims are released
  • how long the assertion or token remains valid
  • how group or role membership maps to permissions
  • whether MFA, device state, network, or risk signals are required
  • what happens when the user is disabled or the session is revoked

If a question says “federation is configured” but access is still too broad, the missing work is usually claims mapping, authorization, conditions, or lifecycle review.

Federation does not remove authorization

After federated sign-in, the target system still decides what the user can do. This is where exam distractors become subtle. A user may authenticate through the correct IdP and still be denied because:

  • the user is not assigned to the application
  • the group claim does not map to the expected role
  • the role lacks the required action
  • a resource policy denies the request
  • a condition such as MFA, device compliance, network, or time is not satisfied
  • an organization-level guardrail overrides the local allow

Treat federation as the beginning of the session path, not the end of access design.

Strong operational controls

Control Why it matters
Centralized lifecycle Joiner, mover, and leaver changes flow to connected applications.
Group or role governance Application access follows approved job function instead of ad hoc assignment.
Strong authentication policy Higher-risk apps can require stronger factors or contextual controls.
Assertion and token lifetime Stolen or stale credentials have a shorter usefulness window.
Audit correlation Sign-in events and application actions can be tied to the same identity.
Break-glass accounts Emergency access remains possible when the federation path fails.

Break-glass accounts are not a reason to keep unmanaged administrator accounts everywhere. They should be rare, monitored, protected, and tested.

Common traps

Trap Better reasoning
“SSO means every user can access every app.” SSO centralizes authentication; app assignment and authorization still matter.
“Federation eliminates local policy.” The relying service still evaluates roles, claims, scopes, and permissions.
“Copy users into every application for resilience.” Prefer central lifecycle with controlled emergency access and documented fallback.
“Use the same broad group for all cloud admins.” Map groups to least-privilege roles and review membership.
“Long sessions are more convenient, so they are better.” Session lifetime should match risk, task duration, and revocation needs.

Sample Exam Question

A company uses a corporate identity provider for cloud administrator sign-in. A former employee still has access to one SaaS application because that application maintains a separate local account. Which improvement best addresses the root problem?

A. Increase the local account password length
B. Federate the SaaS application with the central identity provider and tie access to lifecycle deprovisioning
C. Disable MFA for the SaaS app to simplify access
D. Create a shared admin account for offboarding

Best answer: B. The issue is identity lifecycle drift caused by local account sprawl. Federation plus centralized deprovisioning reduces the chance that a disabled worker keeps access in a separate system.

Revised on Friday, September 11, 2026