Identity & Access Management (OAuth 2.0, OIDC, SAML, Kerberos, MFA & Zero Trust)

Deep dive into Identity and Access Management (IAM): AAA framework, MFA and FIDO2/WebAuthn passkeys, access control models (DAC, MAC, RBAC, ABAC), Kerberos protocol flow, SAML 2.0 vs OAuth 2.0 vs OIDC, RADIUS vs TACACS+, Privileged Access Management (PAM), and NIST SP 800-207 Zero Trust Architecture.

Level
intermediate
Languages
en → en
Cards
32
Markdown
Identity & Access Management (OAuth 2.0, OIDC, SAML, Kerberos, MFA & Zero Trust) (.md)

Flashcards (32)

#Front (Term / Question)Back (Definition / Answer)ExampleMultiple-Choice Options
1What are the five categories of authentication factors, and what constitutes true Multi-Factor Authentication (MFA)?1. Something you know (password, PIN); 2. Something you have (smart card, YubiKey, authenticator app); 3. Something you are (fingerprint, iris, facial biometrics); 4. Somewhere you are (geolocation/IP subnet); 5. Something you do (typing cadence, gait). True MFA requires combining factors from AT LEAST TWO DIFFERENT categories (entering a password + a PIN is single-factor two-step, not MFA).
2What are the core security controls of Privileged Access Management (PAM), including Just-In-Time (JIT) access and Credential Vaulting?PAM secures administrative/root accounts through: 1. Credential Vaulting with automated password rotation after every check-out; 2. Just-In-Time (JIT) ephemeral privilege elevation (zero standing privileges—granting admin rights only for an approved ticket window); 3. Privileged Session Management (bastion/jump host session proxying and keystroke/video recording without exposing raw credentials to the admin).
3Compare Role-Based Access Control (RBAC), Attribute-Based Access Control (ABAC), and Rule-Based Access Control (RuBAC).RBAC assigns permissions to job functions/Groups (e.g., 'Billing_Admin') rather than individuals, preventing privilege creep when roles change. ABAC evaluates dynamic boolean policies combining Subject attributes (role, clearance), Resource attributes (classification), Action, and Environmental context (time, device posture, IP—e.g., XACML/AWS IAM tag policies). RuBAC enforces deterministic system rules (e.g., firewall ACLs or time-of-day restrictions).
4What is an MFA Fatigue (Push Bombing) or Adversary-in-the-Middle (AitM) reverse-proxy phishing attack, and why does FIDO2/WebAuthn defeat both?MFA Fatigue floods a user with push notifications until they press 'Approve'; AitM tools (like Evilginx) proxy real-time login sessions to steal session cookies after TOTP entry. FIDO2/WebAuthn (Passkeys) uses asymmetric public-key cryptography bound strictly to the browser's exact origin domain (`rpId`): the authenticator signs a challenge containing the actual domain, making it cryptographically phishing-resistant.
5Define the Principle of Least Privilege, Separation of Duties, Dual Control, and Job Rotation.Least Privilege: Granting users/processes only the minimum permissions necessary to perform their assigned tasks. Separation of Duties: Splitting critical workflows across multiple roles so no single person can commit and conceal fraud (e.g., developer cannot deploy to prod). Dual Control (Two-Person Integrity): Requiring two authorized individuals to simultaneously approve/act (e.g., two physical keys for an HSM). Job Rotation: Periodically rotating staff across roles to uncover collusion/fraud.
6When should you use the OAuth 2.0 Client Credentials Grant versus Authorization Code + PKCE versus Device Authorization Flow?Client Credentials Grant: Machine-to-machine (M2M) backend microservices or daemons authenticating with their own `client_id` and `client_secret`/private key where no human user is present. Authorization Code + PKCE: All human-facing web, mobile, and single-page applications (Implicit grant is deprecated). Device Authorization Flow: Input-constrained devices like Smart TVs or CLI tools where the user enters a code on a secondary phone/browser.
7What is Break-Glass (Emergency Access) account management, and how should it be monitored?Break-glass accounts are local, non-federated emergency administrator accounts used only when the primary Identity Provider (SSO/MFA/AD) is unavailable during an outage. Their credentials should be split/vaulted (e.g., physical safe or HSM), excluded from normal conditional access blocks, and configured to trigger immediate high-severity SOC alerts the instant any authentication attempt occurs.
8Compare RADIUS and TACACS+ across transport protocols, encryption coverage, AAA separation, and primary use cases.RADIUS: Uses UDP ports 1812/1813 (or 1645/1646), encrypts ONLY the password attribute in the packet (leaving usernames and attributes visible, unless using RadSec over TLS TCP 2083), combines AuthN and AuthZ, and is standard for 802.1X Wi-Fi/VPN user access. TACACS+: Cisco-developed protocol using TCP port 49, encrypts the ENTIRE packet payload, separates AuthN, AuthZ, and Accounting (enabling granular per-command CLI authorization on routers/switches).
9Compare HOTP (RFC 4226) and TOTP (RFC 6238) one-time password algorithms.Both compute an HMAC (typically HMAC-SHA1/256) using a shared secret seed between server and token, then truncate it to a 6–8 digit code. HOTP uses an incrementing event counter (`C`) synchronized on each button press. TOTP replaces the counter with the current Unix timestamp divided by a time step (typically 30 seconds), expiring codes automatically.
10Why does Kerberos require strict NTP clock synchronization (default 5-minute skew tolerance) across all domain hosts?Kerberos embeds encrypted timestamps inside authenticators and tickets to prevent replay attacks without maintaining stateful challenge databases. If a client's system clock drifts more than the configured maximum tolerance (default 5 minutes in Active Directory) from the KDC's clock, the KDC rejects the request with `KRB_AP_ERR_SKEW`.
11Walk through the OAuth 2.0 Authorization Code Grant with PKCE (Proof Key for Code Exchange), and explain what attack PKCE prevents.1. Client generates a high-entropy `code_verifier` and sends its SHA-256 hash (`code_challenge`) in the `/authorize` redirect; 2. User authenticates and browser receives a short-lived `authorization_code`; 3. Client sends `authorization_code` + original plaintext `code_verifier` to `/token`; 4. Authorization server hashes the verifier, checks it matches the initial challenge, and returns tokens. PKCE prevents authorization code interception attacks on public clients (mobile apps/SPAs).
12Compare Network Access Control (NAC) deployment modes: Agent-Based vs Agentless, and In-Band (Inline) vs Out-of-Band.Agent-Based NAC uses a persistent or dissolvable posture agent on the endpoint to inspect OS patch level, EDR status, and disk encryption before granting access; Agentless NAC uses passive profiling (DHCP fingerprinting, MAC OUI, SNMP/Nmap scans) for IoT/printers. Inline NAC sits directly in the data path as a bump-in-the-wire; Out-of-Band NAC communicates with switches/WLCs via 802.1X/RADIUS CoA (Change of Authorization) to reassign VLANs.
13Define False Acceptance Rate (FAR), False Rejection Rate (FRR), and Crossover Error Rate (CER) in biometric authentication.FAR (Type II Error): The rate at which an unauthorized impostor is erroneously granted access (most dangerous for security). FRR (Type I Error): The rate at which a legitimate enrolled user is erroneously denied access (impacts usability). CER (Equal Error Rate / EER): The sensitivity threshold point where `FAR = FRR`; a lower CER indicates a more accurate and reliable biometric system.
14Walk through the 6-step Kerberos (TCP/UDP Port 88) authentication flow using the KDC, AS, TGT, TGS, and Service Ticket.1. Client sends username + timestamp encrypted with password hash to Key Distribution Center (KDC) Authentication Server (AS); 2. AS returns a Ticket Granting Ticket (TGT, encrypted with `krbtgt` key) + session key; 3. Client sends TGT + SPN to Ticket Granting Server (TGS); 4. TGS returns a Service Ticket encrypted with the target server's secret key; 5. Client presents Service Ticket to target server; 6. Server verifies ticket and grants access.
15How do Linux Discretionary Access Control permissions (`rwxr-xr--`, `chmod`, `chown`) and special bits (`SUID`, `SGID`, `Sticky Bit`) work?Standard permissions define Read (4), Write (2), and Execute (1) for Owner, Group, and Others (`rwxr-xr--` = `0754`). SUID (`4000`, `s` on user execute): executes a binary with the file owner's privileges (e.g., `/usr/bin/passwd` running as root; high privilege escalation risk if misconfigured). SGID (`2000`): executes with group privileges or forces new files in a directory to inherit the directory's group. Sticky Bit (`1000`, `t` on `/tmp`): only the file owner or root can delete files inside a shared directory.
16How does SAML 2.0 (Security Assertion Markup Language) enable enterprise Web Single Sign-On (SSO)?SAML 2.0 is an XML-based federation standard involving three parties: Principal (User/Browser), Identity Provider (IdP, e.g., Okta/Entra ID), and Service Provider (SP, e.g., Salesforce). In SP-initiated SSO, the SP redirects the browser to the IdP with a `SAMLRequest`; the user authenticates at the IdP, which returns a digitally signed XML `SAMLResponse` assertion posted via the browser back to the SP's Assertion Consumer Service (ACS).
17What is the difference between an OAuth 2.0 Access Token and a Refresh Token in terms of lifespan, storage, and rotation?Access Token: Short-lived credential (e.g., 5–60 minutes, often a stateless JWT or opaque reference token) sent in the `Authorization: Bearer` header to resource APIs. Refresh Token: Long-lived credential bound to the client and stored securely (e.g., `HttpOnly`, `Secure` cookie or backend vault) used solely at the Authorization Server's `/token` endpoint to obtain a new Access Token via Refresh Token Rotation.
18What are Service Accounts, Managed Identities / IAM Roles, and Workload Identity Federation, and why should static API keys be avoided?Traditional service accounts use long-lived passwords or static API keys that frequently leak in git repositories or environment files. Cloud Managed Identities (Azure), IAM Instance Profiles/IRSA (AWS), and Workload Identity Federation (OIDC token exchange) issue short-lived, automatically rotated STS tokens to workloads without storing any static secrets in code.
19Compare Discretionary Access Control (DAC) and Mandatory Access Control (MAC).DAC: Every resource has an Owner who has full discretion to grant or revoke permissions (ACLs) to other users (e.g., standard NTFS or Unix `chmod`); flexible but vulnerable to Trojan horses inheriting user permissions. MAC: System-enforced access based on security clearances of subjects and sensitivity labels of objects (e.g., Top Secret, SELinux); users cannot alter permissions.
20In Mandatory Access Control, how does the Bell-LaPadula model differ from the Biba integrity model?Bell-LaPadula enforces Confidentiality: 'No Read Up' (Simple Security Property: a Secret user cannot read Top Secret files) and 'No Write Down' (*-Property: a Top Secret user cannot write to a Secret file to prevent data leakage). Biba enforces Integrity: 'No Read Down' (cannot read lower-integrity untrusted data) and 'No Write Up' (cannot corrupt higher-integrity data).
21What is the fundamental difference between OAuth 2.0 and OpenID Connect (OIDC)?OAuth 2.0 (RFC 6749) is strictly a delegated AUTHORIZATION framework: it issues an Access Token allowing a client application to access protected APIs on behalf of a resource owner, but does not define user identity. OpenID Connect (OIDC) is an AUTHENTICATION identity layer built on top of OAuth 2.0 that adds a standardized JSON Web Token (JWT) called an `id_token` containing verified user identity claims (`sub`, `iss`, `aud`, `exp`) and a `/userinfo` endpoint.
22What is Privilege Creep, and how do Identity Governance and Administration (IGA) Access Reviews mitigate it?Privilege Creep occurs when employees accumulate permissions across internal department transfers or promotions without revoking old access rights, resulting in excessive standing privileges. IGA platforms mitigate this via automated Joiner-Mover-Leaver (JML) lifecycle workflows and periodic User Access Reviews (attestation/recertification campaigns) where managers must explicitly re-approve or revoke access.
23What is Kerberos Constrained Delegation vs Unconstrained Delegation in Active Directory, and why is Unconstrained Delegation dangerous?Unconstrained Delegation allows a middle-tier server (e.g., a web server) to impersonate a user to ANY service in the domain; when a user connects, the KDC places a copy of the user's TGT inside the service ticket cached in the server's LSASS memory—if an attacker compromises that server, they can steal Domain Admin TGTs. Constrained Delegation (`S4U2Proxy`) restricts impersonation strictly to specified target SPNs.
24What are Session Management best practices after authentication succeeds?1. Generate a high-entropy random Session ID using a CSPRNG; 2. Regenerate the Session ID immediately upon login or privilege change (preventing Session Fixation); 3. Transmit session tokens exclusively in cookies with `HttpOnly` (blocks XSS `document.cookie` theft), `Secure` (HTTPS only), and `SameSite=Lax/Strict` attributes; 4. Enforce both idle and absolute session timeouts and server-side invalidation on logout.
25What is IEEE 802.1X Port-Based Network Access Control, and what are the roles of the Supplicant, Authenticator, and Authentication Server?802.1X blocks network access on a switch port or Wi-Fi SSID until identity is verified. Supplicant: Client software/OS requesting access. Authenticator: Network switch or wireless access point enforcing port blocking (forwarding only EAPOL frames on the unauthorized port). Authentication Server: RADIUS server validating credentials via EAP (Extensible Authentication Protocol) and returning an `Accept` or `Reject` (plus dynamic VLAN ID).
26Define the four components of the IAAA (Identification, Authentication, Authorization, and Accounting) security framework.Identification: Claiming an identity (e.g., entering a username or email). Authentication (AuthN): Proving that identity claim using one or more factors. Authorization (AuthZ): Determining what actions or resources the authenticated identity is permitted to access. Accounting (Auditing): Logging and tracking user actions, session duration, and resource consumption for non-repudiation and forensics.
27Compare EAP-TLS, PEAP, EAP-TTLS, and EAP-FAST in 802.1X wireless and wired authentication.EAP-TLS: Most secure; requires X.509 digital certificates installed on BOTH the RADIUS server and every client device (mutual TLS). PEAP (Protected EAP) and EAP-TTLS: Require a certificate ONLY on the RADIUS server to build an outer TLS tunnel, then authenticate the user inside the tunnel using MS-CHAPv2 or PAP/passwords. EAP-FAST: Uses Protected Access Credentials (PACs) instead of certificates.
28What is Federated Identity Management, and what is the difference between Home Realm Discovery and Identity Brokering?Federation links identity management across distinct security domains via trust relationships (SAML/OIDC/WS-Federation). Home Realm Discovery (HRD) determines which Identity Provider a user belongs to based on their email domain (`@corpA.com` vs `@corpB.com`). An Identity Broker acts as an intermediary hub translating between multiple IdPs and multiple Service Providers.
29Explain how SCIM (System for Cross-domain Identity Management) complements SAML/OIDC in enterprise identity lifecycle management.While SAML and OIDC handle real-time user authentication (SSO), they do not automatically deprovision accounts or sync group memberships before a user logs in. SCIM 2.0 (RFC 7644) provides a standardized REST/JSON API for automated user provisioning, group synchronization, and immediate account deprovisioning between an IdP (e.g., Okta, Entra ID) and SaaS applications.
30Compare LDAP (Port 389) and LDAPS (Port 636) / StartTLS, and explain what an X.500 Distinguished Name (DN) looks like.LDAP (Lightweight Directory Access Protocol) queries and modifies hierarchical directory services (like Active Directory or OpenLDAP) over cleartext TCP/UDP 389 unless upgraded via StartTLS; LDAPS wraps LDAP over TLS on TCP 636. A Distinguished Name (DN) uniquely identifies an entry from leaf to root, e.g., `CN=Alice Smith,OU=Security,DC=corp,DC=example,DC=com`.
31What is Discretionary vs Mandatory vs Risk-Based (Adaptive) Authentication?Adaptive / Risk-Based Authentication continuously evaluates contextual signals during login and throughout an active session—such as impossible travel (logging in from New York and Tokyo within 10 minutes), unfamiliar device fingerprint, Tor/VPN exit node IP, or anomalous behavioral biometrics—to dynamically step up authentication (triggering FIDO2 MFA) or terminate the session.
32What are the core tenets and three logical components of Zero Trust Architecture under NIST SP 800-207?Zero Trust assumes no implicit trust based on network location ('Never Trust, Always Verify'; assume breach; verify explicitly per-session/per-request). NIST SP 800-207 defines the Control Plane—consisting of the Policy Engine (PE, evaluates trust algorithms/signals) and Policy Administrator (PA, issues session tokens/commands)—and the Data Plane, which is the Policy Enforcement Point (PEP, proxy/gateway allowing or terminating traffic).