What is PKI? And how it secures just about everything online
- 14 June, 2019 20:00
Public key infrastructure (PKI) is a catch-all term for everything used to establish and manage public key encryption, one of the most common forms of internet encryption. It is baked into every web browser in use today to secure traffic across the public internet, but organizations can also deploy it to secure their internal communications.
The most crucial concept involved in PKI is, as its name implies, the public cryptographic keys that are at its core. These keys not only are part of the encryption process, but they help authenticate the identity of the communicating parties.
Why is PKI important? It's because this combination of encryption and authentication makes trustworthy online communication possible.
How does PKI work?
The most important concepts to understand in order to grasp how PKI works are keys and certificates. A key, as we've already noted, is a long string of bits — a number, in other words — that's used to encrypt data. For instance, if you used the ancient and simple Caesar cipher with a cryptographic key of 3, that would mean that every letter in your message is replaced by one three letters later in the alphabet — A becomes D, B becomes E, and so forth. In order to decode its message, your recipient would need know not only that you were using the Caesar cipher but that your key was 3.
Obviously the mathematics behind modern encryption is much more complicated than this. One of the ways it's different gets around a somewhat obvious problem with the Caesar cipher: you have to somehow let your recipient know the key used to encode the encrypted message. PKI gets its name because each participant in a secured communications channel has two keys. There's a public key, which you can tell to anyone who asks and is used to encode a message sent to you, and a private key, which you keep secret and use to decrypt the message when you receive it. The two keys are related by a complex mathematical formula that would be difficult to derive from brute force. If you want to get into the weeds on this form of encryption, known as asymmetrical cryptography, HowStuffWorks has a good deep dive.
So that covers how data is encrypted within a public key infrastructure. But remember, PKI is widely used because, in addition to encrypting messages, it also lets you know that the person with whom you're exchanging encrypted messages is who they say they are. And that's where certificates come in.
What are PKI certificates?
PKI certificates are documents that act as digital passports, assigned to any entity that wants to participate in a PKI-secured conversation. They can include quite a bit of data. One of the most important pieces of information a certificate includes is the entity's public key: the certificate is the mechanism by which that key is shared. But there's also the authentication piece. A certificate includes an attestation from a trusted source that the entity is who they claim to be. That trusted source is generally known as a certificate authority, and we'll go into more detail on that in a moment.
With these concepts under our belt, these are the elements that go into PKI.
- A certificate authority (CA), which issues digital certificates, signs them with its own public key, and stores them for reference.
- A registration authority, which verifies the identities of those requesting digital certificates. (We'll get into the ins and outs of how that works in the next section.) A certificate authority can act as its own registration authority or can use a third party to do so.
- A certificate database that stores both the certificates and metadata about them—most importantly, the period of time for which the certificate is valid.
- A certificate policy outlining the PKI's procedures, which allows outsiders to judge how trustworthy the PKI is.
How to get a PKI certificate
From the description of those components, you can see that trust is at the center of any PKI infrastructure. One of the things I'm doing when I send you my digital certificate is trying to get you to trust that I am who I say I am—and the certificate helps by having a trusted third party vouch for me.
To understand how that works in practice, let's consider for a moment the most widespread public key infrastructure system out there: the TLS/SSL protocol, which secures just about all encrypted HTTP communication.
In order to offer TLS-encrypted communication, the owner of a website needs to obtain a certificate from a certificate authority, as we've already discussed. There are a wide range of vendors who've set themselves up as CAs, and before they issue one to you, they require that you prove your ownership of your website in some way. For instance, if you're trying to buy an SSL certificate for a website at example.com, you may need to send the CA an email from email@example.com, which is an address that would be restricted to someone with management authority over that domain name. Once you acquire the certificate, you can upload it to your web server.
The next obvious question, of course, is how you know you can trust the CA: after all, in the case of TLS, there's no centralized body in charge of the standard and anyone can set themselves up as a certificate authority. A bad actor in that role could wreak havoc. In practice, OS and browser makers like Apple, Microsoft, and Mozilla are the de facto gatekeepers here, maintaining lists of trusted CAs and blacklisting those who slip. The decisions on which CAs to trust have high stakes, as a 2017 showdown between Google and Symantec over what Google felt were Symantec's lax standards made clear.
An SSL certificate is probably the most common type of PKI certificate you would encounter in the wild. But as we'll discuss in a moment, it isn't the only kind. Other PKIs will have different standards for issuing certificates, but the important thing to keep in mind is that any PKI system must have some method by which CAs can authenticate users, and that all participants in the PKI system trust that method. SSL/TLS uses what's called a chain of trust, in which users eventually must choose to trust a root certificate-granting authority; an alternate scheme, outlined by Finjan Cybersecurity, is the web of trust, in which users' certificates are signed by other users. A web of trust system is better suited to self-contained networks or organizations, or small communities of users.
What is PKI used for?
SSL may be the most widespread implementation of PKI, but it certainly isn't the only one. This Experts Exchange thread has a great list of real-world PKI applications, including:
- Providing a recovery key for an encrypted hard drive
- Securing internal communications with database servers
- Signing documents
- Securing local networks — PKI capacities are built into Microsoft's Active Directory, for instance, and can work with physical keycards that store digital certificates to ensure that users are who they say they are.
- Secure messaging — The Signal protocol uses PKI, for instance
- Email encryption
Why do we need PKI for secure email?
PKI is great for securing email for the same reason that it's great for securing web traffic: because data flowing over the open internet can be easily intercepted and read if it isn't encrypted, and because it can be difficult to trust that a sender is who they claim to be if there isn't some way to authenticate their identity. As we've seen, establishing near-universal PKI for web traffic has been relatively easy because most of the necessary infrastructure is built into web browsers and servers. Email is accessed through more heterogenous clients, which makes things a bit trickier.
One of the oldest and best-established PKI systems for securing email is S/MIME; there's also PGP (Pretty Good Privacy), which uses the web of trust model we discussed above. Support for these kinds of email protections are built into clients like Microsoft Outlook. The rise of web-based email in recent years has seen a step backwards in this regard. Gmail, for instance, only supports S/MIME in paid G Suite accounts, not free accounts.
This has been a high-level introduction to the concepts around PKI. SmallStep, an open source identity infrastructure company, has a wonderfully long and detailed article called "Everything you should know about certificates and PKI but are too afraid to ask" that can take you much, much more in depth. Among other things, SmallStep takes you through the process of actually issuing certificates, so you can see what they contain.
If you're looking for a way to set up a public key infrastructure and play with it to understand some of the basic concepts, this tutorial on the Gentoo Wiki explains how to do it on Linux. If you want to see how you'd build a certificate authority for an in-house PKI, HashiCorp has a tutorial on how to do that with their Vault engine, which should demonstrate the concepts.