What is a PKI? Terms explained
What is a PKI?
A PKI (public key infrastructure) is, according to the NIST glossary, a framework established to issue, maintain and revoke public key certificates. RFC 4949 describes it as the hardware, software, people, policies and procedures needed to create, manage, store, distribute and revoke digital certificates based on asymmetric cryptography.
A PKI is therefore not a single piece of software, but the interplay of technology, rules and the people who run it. RFC 4949 names three core functions:
- register users and issue their certificates,
- revoke certificates when required,
- archive the data needed to validate certificates much later.
What problem does a PKI solve?
Anyone using a public key has to be sure that the matching private key belongs to the right counterpart, a person or a system. According to RFC 5280, this confidence comes from certificates: data structures that bind public key values to subjects.
The binding holds because a trusted certificate authority (CA) digitally signs each certificate. How a server sends its certificate when a connection starts, and how you look at it, is shown in Checking an SSL/TLS certificate.
What are a public key and a private key?
A key pair has two parts. According to RFC 4949, the public key is the publicly disclosable component of a pair of cryptographic keys used for asymmetric cryptography, and the private key is the secret component.
By RFC 4949 (quoting X.509), the private key is the key of a user’s key pair that is known only by that user. Only the subject’s public key goes into the certificate.
What does a certificate contain?
The usual format is X.509: RFC 5280 describes X.509 as the standard that defines a certificate format. According to RFC 5280, section 4.1.2, every certificate contains:
- the names of the subject and the issuer,
- the public key of the subject,
- a validity period,
- a version number and a serial number.
The issuing CA signs this information. With that signature, RFC 5280 says, the CA certifies the validity of the information and in particular the binding between the public key and the subject of the certificate.
Are there different kinds of certificates?
RFC 5280 covers two classes: CA certificates and end entity certificates. A CA certificate belongs to a certificate authority, an end entity certificate to the user or system at the end of the chain, such as a server.
Who takes part in a PKI?
RFC 5280 describes a simplified model with five components:
- End entity: a user of PKI certificates or an end user system that is the subject of a certificate.
- CA (certification authority): the authority that issues certificates. Its role is covered in What is a certificate authority (CA)?.
- RA (registration authority): an optional system to which a CA delegates certain management functions.
- CRL issuer: a system that generates and signs CRLs.
- Repository: a system or collection of systems that stores certificates and CRLs and distributes them to end entities.
What does a registration authority do?
According to RFC 4949, an RA signs neither certificates nor CRLs. It records or verifies information, above all the identity of applicants, that a CA needs in order to issue certificates. A CA can perform all management functions itself, or hand secondary functions to an RA and keep the signing (RFC 4949).
How does trust in a certificate come about?
A client knows only a limited number of CA keys in advance. RFC 5280 therefore describes chains, called certification paths: a certificate of the key owner signed by one CA, plus zero or more further CA certificates signed by other CAs.
According to RFC 5280, such chains are needed because a client starts with only that limited number of assured CA public keys. How a missing level shows up is described in nginx: incomplete certificate chain. If the top of the chain is unknown to the client, Internal CA: unknown root helps.
What is a trust anchor?
RFC 4949 defines a trust anchor as an established point of trust from which a client begins the validation of a certification path. According to RFC 5280, choosing it is a matter of policy: it can be the top CA of a hierarchy, or the CA that issued the verifier’s own certificate.
What is a root certificate?
By RFC 4949, the root certificate is the self-signed certificate at the top of a certification hierarchy. Its signature is made with the private key that belongs to the public key in the certificate (RFC 5280).
According to RFC 4949, the root’s public key is distributed in a way that does not depend on the PKI’s certification relationships, that is, outside the PKI itself. How hierarchies are built is described in Building a PKI.
How do I assemble unordered certificate files into a chain?
The PKI Toolkit takes a pile of PEM, CRT and CER files in any order and returns the correctly ordered chain as fullchain.pem. According to its tool page, it runs entirely in the browser, without upload.
How long is a certificate valid, and how is it revoked?
Every certificate has a validity period with a start (notBefore) and an end (notAfter). RFC 5280 defines it as the interval during which the CA warrants that it will maintain information about the status of the certificate.
A certificate can also become invalid before it ends. RFC 5280 lists a change of name and the compromise or suspected compromise of the private key among the reasons, and says the CA then needs to revoke the certificate. RFC 5280 names, among others, OCSP and certificate revocation lists (CRLs).
What is a certificate revocation list (CRL)?
According to RFC 5280, a CRL is a time-stamped list identifying revoked certificates, signed by a CA or CRL issuer and made available in a public repository. Each revoked certificate appears in it with its serial number.
The weakness: the time granularity of revocation is limited to the CRL issue period (RFC 5280). Where a certificate announces its CRLs is given by the “CRL Distribution Points” extension.
What is OCSP?
OCSP (Online Certificate Status Protocol) determines the current status of a certificate without requiring CRLs (RFC 6960). The answer is “good”, “revoked” or “unknown” and must be digitally signed.
What is a PKI used for?
The NIST glossary names encryption, digital signature and authentication as the purposes certificates support. RFC 5280 shows purposes in the “Extended Key Usage” extension, which states what the certified key may be used for:
- TLS server authentication (
serverAuth), - TLS client authentication (
clientAuth), - signing of downloadable executable code (
codeSigning), - email protection (
emailProtection).
If the extension is present, RFC 5280 says the certificate must be used only for one of the purposes indicated. The Mozilla Root Store Policy names TLS servers and S/MIME email users as the purposes of its root store.
Where do I read on?
- What is a certificate authority (CA)?: issuing, signing, revoking, public and private.
- Building a PKI: root CA, intermediate CA, Windows, smartcard: structure and levels.
- Setting up your own certificate authority (CA) on Windows: the steps with Active Directory Certificate Services.
- Managed PKI or your own PKI: operating models and providers.
Sources
- RFC 5280: Internet X.509 Public Key Infrastructure Certificate and CRL Profile, retrieved on 19 November 2026
- RFC 4949: Internet Security Glossary, Version 2, retrieved on 19 November 2026
- RFC 6960: X.509 Internet PKI Online Certificate Status Protocol - OCSP, retrieved on 19 November 2026
- NIST CSRC Glossary: public key infrastructure (PKI), retrieved on 19 November 2026
- Mozilla Root Store Policy, retrieved on 19 November 2026