Security in depth · Unit 24 · Lesson 6 of 14
Certificates: who gets to vouch for whom?
Understand public keys, issuers, validation, and the difference between encryption and identity.
Helpful before thisActive Directory
After this lesson you can
- explain what a certificate binds
- identify issuer and validation dependencies
- distinguish proof of key possession from authorization
A service presents a digital certificate. It is tempting to read the padlock as “this service is good.” The certificate makes a narrower claim: it binds an identity or name to a public key under a trust system.
Certificate chain: A sequence of signed certificates connecting an identity or key to an accepted trust anchor.
The certificate is not the private key
A certificate includes a public key and signed information about its subject, issuer, and permitted uses. The associated private key must remain protected. In a protocol, a party can demonstrate possession of that private key; merely copying a public certificate does not provide that ability.
A certificate chain connects a leaf certificate through issuers to a trust anchor the verifier accepts. The trust anchor is a local policy choice, not a universal vote that every issuer is trustworthy.
Validation has several questions
A relying party checks the relevant chain, signatures, time validity, intended identity, and usage constraints. Revocation handling depends on the application and deployment. A valid signature alone is not the entire validation process.
In an enterprise, certificate enrollment is itself an authorization problem: who may request which identity, for what purpose, and with whose approval? Templates and enrollment rights can influence which assertions an issuer signs.
Keep the claim narrow
A certificate used for client authentication does not automatically grant administration. The receiving service maps the authenticated identity to permissions. A certificate used for TLS helps authenticate the connection endpoint; it does not prove the application has no vulnerabilities or that every statement it makes is true.
Operational quality matters: track owners, expiration, renewal, protected key storage, and recovery. An expired certificate can cause an outage, while a poorly controlled issuer can damage the integrity of the whole trust system.
EXPLORE THE CONCEPT
Three checks, three meanings
Explore what each observation establishes.
The certificate signature verifies
The signed data matches the issuer’s signature. Identity, lifetime, usage, and trust-chain checks still matter.
The party proves possession of the private key
It demonstrates control of the key in the protocol. Authorization remains a separate decision.
The certificate has expired
The validity window has ended. Renewal and deployment need to be part of the service’s operating process.
A simplified learning model. It connects to no systems and uses no real data.
Turn the idea into a decision
Ask both “who may receive this credential?” and “what will accept it?” Trust is a relationship between issuance and use.
Worked review: the certificate belongs to the wrong name
The fictional Elm archive is moving its HTTPS service. The client uses normal DNS-name validation, without a special identity override. Its configured destination is archive.elm.example.
C1: The presented leaf certificate’s only DNS subject alternative name is old-archive.elm.example. There are no wildcards or other matching identifiers.
C2: Chain signatures, validity, and server-authentication usage meet the client’s policy. The chain reaches an already accepted trust anchor. All required checks other than the expected-name comparison pass.
C3: The server proves possession of the associated private key during the protocol. The proposed change note says: “Accept because the issuer is trusted.”
C4: The archive owner approved the new service name. Application authorization permits ResearchReaders to read documents and reserves deletion for ArchiveMaintainers.
PredictDo C2 and C3 justify accepting the service as archive.elm.example?
No. C1 fails the expected-name comparison even though the chain and key-possession checks pass. These checks establish different facts. Adding the same issuer to another trust store would not supply the missing name.
The narrow correction is to obtain and deploy a certificate for the owner-approved destination through the established issuance process, then verify the intended client’s complete validation. Redirecting users to a different name would require its own approved service design; it should not silently redefine the identity the client intended to contact.
Issuance policy and use policy meet
In AD Certificate Services, templates help define certificate properties and enrollment permissions. A review asks which authenticated requesters may enroll, which subject information is accepted, which uses are allowed, and who may change those rules. Template permission alone is not proof that a request was issued or accepted by every application.
Possession of the public certificate is not possession of its private key. Conversely, proving control of the private key does not prove that a human or service deserves every operation. Even after endpoint authentication succeeds, C4 still separates reading from deletion.
Track the deployed certificate and protected key separately through renewal. Issuing a replacement does not prove clients receive it; a deployment may still serve the old certificate. Acceptance should therefore include the expected name, validation outcome on the intended client, and preserved application permissions. Record revocation behavior according to that client’s actual policy instead of assuming every verifier handles unavailable status information identically.
Terms you met
Check yourself
No timer. No penalties. Read the explanation and try again whenever you like.
This lesson’s questions have changed. Your reading progress is saved; review the updated questions.
-
C3 shows key-possession proof. What would merely copying the public certificate provide?
Show the answer
Correct answer: The public key and signed claims, without the private-key proof. Copying public data does not establish control of the corresponding secret key in the protocol.
-
Which next step addresses the expected-name failure in C1-C3?
Show the answer
Correct answer: Obtain and deploy a certificate for the approved destination, then verify the intended client’s complete validation. This addresses the missing DNS identity match while preserving the separate chain, time, usage and key-possession requirements.
-
After the name is corrected, a ResearchReaders member requests deletion. What follows from C4?
Show the answer
Correct answer: The application should deny deletion despite successful endpoint authentication. C4 reserves that operation for another role. Authenticating a connection does not authorize every application action.
-
The issuer records successful renewal. What further evidence is needed before closing the migration?
Show the answer
Correct answer: The intended client receives and validates the correct certificate, with application permissions preserved. This connects issuance to deployment and use while keeping authorization separate.
Try it
- WritePrepare a trust review for C1-C4. List issuance, endpoint validation, and resource authorization as separate decisions. Name the missing check that blocks the proposed migration, give a narrow correction, and explain why a copied public certificate or added trust anchor would not resolve it.
References
- RFC 5280: Certificate and CRL profile
- RFC 8446: TLS 1.3
- Wikipedia: Public key infrastructure · General background; use the technical sources for implementation details.
- RFC 9525: service identity in TLS
- Microsoft: certificate template concepts