Password Policies and Authentication Alternatives
A password is just one way to prove who you are, and the weakest one in the set. This lesson puts strong password policy next to its alternatives — SSH, certificates, MFA, biometrics — and then shows how AAA centralises authentication, authorisation, and accounting across every device. The exam asks both halves: what makes a strong credential, and how the AAA architecture (TACACS+ and RADIUS) manages it at scale.
- Write a strong password policy: length, character classes, and rotation.
- Apply service password-encryption, enable secret, and login local.
- Justify SSH over Telnet on a management plane.
- Explain MFA, certificates, and biometrics as authentication alternatives.
- Break down AAA into authentication, authorisation, and accounting.
- Compare TACACS+ (TCP 49) and RADIUS (UDP 1812/1813).
A strong password policy
A password policy is the organisation’s contract about credentials. The modern guidance (echoing NIST SP 800-63B) has moved away from “change every 90 days” as the headline rule and toward the things that actually stop attackers:
- Length wins. Long passphrases (12+ characters) defeat brute force better than short complex ones. “Correct-horse-battery-staple” beats “p@5s-W0rd”.
- Character diversity. Mix upper, lower, digits, and symbols so the passphrase is not in any word list.
- No reuse. The same credential must not appear across systems — a breach anywhere leaks credentials everywhere else.
- Rotation on suspicion only. Change after a suspected compromise or a breach rather than on a rigid calendar; frequent forced rotation encourages weak passwords written on sticky notes.
- Multi-factor everywhere else. For management access and admin accounts, policy demands a second factor (Section 5), not a better single password.
Passwords on the device: encryption and enable secret
Lesson 4 covered the mechanics; here is the policy view. On a Cisco device, enable secret protects privileged EXEC mode with a one-way hash, while enable password stores cleartext or weak type-7 obfuscation. The rule: always use the secret; never the password. For every other stored password (console, VTY, local users), service password-encryption applies type-7 scrambling so the running config does not display readable secrets:
R1(config)# enable secret Str0ngSecret! R1(config)# service password-encryption R1(config)#
service password-encryption does not truly encrypt — type 7 is a fixed, reversible
obfuscation that any free tool can crack in seconds. Use it to deter shoulder-surfers, not as a
security control. The hash inside enable secret is the real protection; newer IOS can use
SHA-256 via enable algorithm-type sha256 secret instead of legacy MD5.
Local user accounts: login local
Instead of giving every line its own password, IOS can keep a local database of usernames and passwords. The VTY and console lines then use login local to authenticate against that database, which gives each administrator a name for audit trails and lets individual accounts be revoked without changing a line password:
R1(config)# username admin privilege 15 secret Password! R1(config)# username ops privilege 5 secret Ops-2026! R1(config)# line vty 0 4 R1(config-line)# login local R1(config-line)# transport input ssh
Each username carries a privilege level (0-15). Privilege 15 is full control; lower levels limit commands, which is a crude authorisation mechanism. Local accounts are fine for a handful of devices, but they do not scale — every router keeps its own copy of every credential, so hiring and firing means touching each device. That scaling problem is exactly what AAA (Section 8) solves.
SSH vs Telnet
Both protocols open a remote command shell; the difference is the transport. Telnet sends every byte — including the password — in cleartext over TCP 23. Anyone sniffing the path (or any switch with a port mirror) can replay the session. SSH (Secure Shell) encrypts the entire session over TCP 22 and authenticates the server with a key pair first, so sniffing yields ciphertext and the password never travels in the clear.
| Property | Telnet | SSH |
|---|---|---|
| Port | TCP 23 | TCP 22 |
| Encryption | None — cleartext | Full-session encryption |
| Server identity | Not verified | Verified via host key |
| Modern verdict | Deprecated; disable with transport input ssh | Required standard |
transport input ssh on line vty 0 4 — allowing SSH only. Any question
showing Telnet on a management line is asking for this fixation.
Multi-factor authentication
MFA requires two or more independent factors before access succeeds. The three factor classes are: something you know (password, PIN), something you have (smartphone app, hardware token, smart card), and something you are (fingerprint, face). Two different factors — say a password plus a one-time code from an app — defeat the two most common attacks at once: if the password leaks, the attacker still lacks the token, and if the token is duplicated, they still lack the password.
MFA is cheap now, and it is the difference between a password breach being a nuisance and a password breach being a disaster. For admin access to infrastructure (the CCNA/CyberOps audience sit right here), MFA is baseline practice, not a nice-to-have.
Certificates
A digital certificate is a credential that binds an identity (a person, a device, a server) to a public key, signed by a trusted authority. Authentication with certificates works by proving possession of the matching private key during a challenge — the holder never sends the private key itself:
- Certificate Authority (CA) signs and issues the certificate.
- Identity verification happens when the peer demonstrates it owns the private key corresponding to the public key in the certificate.
- Revocation (CRL/OCSP) lets an authority say “this certificate is no longer valid” if the private key leaks or the device is decommissioned.
Certificates are ideal for device-to-device authentication because there is no shared secret to copy around each device holds its own key pair, and they scale with CA-managed issuance and revocation. On the exam, remember whose job the CA does (authoritative identity), and that the private key must never leave its device.
Biometrics
Biometrics authenticate on something you are: fingerprint, iris, face, or voice. They are convenient and hard to share — you cannot forget your fingerprint the way you lose a password — but they are not magical:
- Biometric samples must be captured, stored, and compared by hardware and software that itself needs protecting (spoofing, replay, template theft).
- Failure modes are irreversible: a stolen password can be changed, a stolen fingerprint cannot.
- The practical use in enterprise access is MFA layering: fingerprint on the laptop plus a password or token, rather than biometrics alone.
AAA: authentication, authorisation, accounting
AAA centralises access control in three distinct services:
| Service | Job | Asks... |
|---|---|---|
| Authentication | Verifies who the user is | “Are you really admin?” |
| Authorisation | Decides what the user may do | “Are you allowed to run configure terminal?” |
| Accounting | Logs what the user did | “What commands did you run, and when?” |
With AAA, the device phones home to a central server on login. Authentication uses the server’s
verdict; authorisation returns command- and service-level permissions; accounting ships start/stop
records of sessions and commands to a log. The benefit over local accounts is governance: one database,
one revocation path, and a complete audit trail. IOS activates it with
aaa new-model, then configures the server and methods:
R1(config)# aaa new-model R1(config)# aaa authentication login default group tacacs+ local R1(config)# aaa authorization exec default group tacacs+ R1(config)# aaa accounting exec default start-stop group tacacs+ R1(config)# tacacs server TAC1 R1(config-server-tacacs)# address ipv4 192.168.1.50 R1(config-server-tacacs)# key TacacsKey123 R1(config-server-tacacs)# exit
The catch-all fallback phrase “group tacacs+ local” means: try TACACS+ first, and if the server is unreachable, fall back to local accounts — the safe default for keeping management access alive during an outage.
TACACS+ vs RADIUS
Two server protocols dominate AAA, and the exam asks for their differences constantly:
| Property | TACACS+ | RADIUS |
|---|---|---|
| Transport & port | TCP 49 | UDP 1812/1813 |
| Encryption | Encrypts the whole packet body | Encrypts only the password (rest of packet readable) |
| Authorisation | Separate authorisation step, records per-command levels | Merges authorisation with authentication response (attributes) |
| Accounting | Separate, flexible accounting subsystem | Built-in accounting (start/stop/update) |
| Typical home | Cisco device management | Network access (wireless, VPN, wired 802.1X) and general services |
The question that decides which one a scenario wants: is this device administration or network access? Device administration (login to the router, per-command permissions) leans TACACS+. Network access (WPA2-Enterprise wireless, VPN users, 802.1X) leans RADIUS. Neither is exclusive, but the blueprint tests the association.
The vocabulary you need
Every term below appears in the lesson text above. Select one to read its definition, then use the quiz at the end to check what stuck.
Review
Check what you have learned with these review questions.
Sources & references
- Cisco. CCNA (200-301) Exam Topics. cisco.com.
- IETF. Request for Comments (RFCs). rfc-editor.org.
- IANA. Internet Assigned Numbers Authority registries. iana.org.
Target exams: CCNA 200-301 v1.1 · CyberOps Associate 200-201. Last content verification: 2026-10-11.