top of page

What is OAuth Internet Protocol?

OAuth allows notifying a resource provider (e.g. Facebook) that the resource owner (e.g. you) grants permission to a third-party (e.g. a Facebook Application) access to their information (e.g. the list of your friends).

If you read it stated as plainly, I would understand your confusion. So let’s go with a concrete example: joining yet another social network!

Say you have an existing GMail account. You decide to join LinkedIn. Adding all of your many, many friends manually is tiresome and error-prone. You might get fed up half-way or insert typos in their e-mail address for invitation. So you might be tempted not to create an account after all.

Facing this situation, LinkedIn has the Good Idea(TM) to write a program that adds your list of friends automatically because computers are far more efficient and effective at tiresome and error prone tasks. Since joining the network is now so easy, there is no way you would refuse such an offer, now would you?

Without an API for exchanging this list of contacts, you would have to give LinkedIn the username and password to your GMail account, thereby giving them too much power.

This is where OAuth comes in. If your GMail supports the OAuth protocol, then LinkedIn can ask you to authorise them to access your GMail list of contacts.

OAuth allows for:

  1. Different access levels: read-only VS read-write. This allows you to grant access to your user list or a bi-directional access to automatically synchronize your new LinkedIn friends to your GMail contacts.

  2. Access granularity: you can decide to grant access to only your contact information (username, e-mail, date of birth, etc.) or to your entire list of friends, calendar and what not.

  3. It allows you to manage access from the resource provider’s application.

  4. If the third-party application does not provide mechanism for cancelling access, you would be stuck with them having access to your information. With OAuth, there is provision for revoking access at any time.

The Protocol becomes easier when you know the involved parties. Basically there are three parties involved: oAuth Provider, oAuth Client and Owner.

  1. oAuth Client (Application Which wants to access your credential)

  2. oAuth Provider (eg. facebook, twitter…)

  3. Owner (the person with facebook,twitter.. account )

Since the beginning of distributed personal computer networks, one of the toughest computer security nuts to crack has been to provide a seamless, single sign-on (SSO) access experience among multiple computers, each of which require unrelated logon accounts to access their services and content.

Although still not fully realized across the entire internet, myriad, completely unrelated websites can now be accessed using a single physical sign-on. You can use your password, phone, digital certificate, biometric identity, two-factor authentication (2FA) or multi-factor authentication (MFA) SSO solution to log onto one place, and not have to put in another access credential all day to access a bunch of others. We have OAuth to thank for much of it.

What is OAuth??

OAuth is an open-standard authorization protocol or framework that describes how unrelated servers and services can safely allow authenticated access to their assets without actually sharing the initial, related, single logon credential. In authentication parlance, this is known as secure, third-party, user-agent, delegated authorization.

Created and strongly supported from the start by Twitter, Google and other companies, OAuth was released as an open standard in 2010 as RFC 5489, and quickly became widely adopted. Over the next two years, it underwent substantial revision, and version 2.0 of OAuth, was released in 2012 as RFC 6749. Even though version 2.0 was widely criticized for multiple reasons covered below, it gained even more popularity. Today, you can add Amazon, Facebook, Instagram, LinkedIn, Microsoft, Netflix, Paypal and a list of other Internet Who’s as adopters.

OAuth examples:

The simplest example of OAuth is when you go to log onto a website and it offers one or more opportunities to log on using another website’s/service’s logon. You then click on the button linked to the other website, the other website authenticates you, and the website you were originally connecting to logs you on itself afterward using permission gained from the second website.

Another common example OAuth scenario could be a user sending cloud-stored files to another user via email, when the cloud storage and email systems are otherwise unrelated other than supporting the OAuth framework (e.g., Google Gmail and Microsoft OneDrive). When the end-user attaches the files to their email and browses to select the files to attach,

OAuth could be used behind the scenes to allow the email system to seamlessly authenticate and browse to the protected files without requiring a second logon to the file storage system. Another example, one given in the OAuth 2.0 RFC, is an end-user using a third-party printing service to print picture files stored on an unrelated web server.

In all cases, two or more services are being used for one transaction by the end-user, and every end-user would appreciate not being asked to log in a second time for what they feel is a single transaction. For OAuth to work, the end-user’s client software (e.g., a browser), the services involved and authentication provider must support the right version of OAuth (1.0 versus 2.0).

OAuth explained:

When trying to understand OAuth, it can be helpful to remember that OAuth scenarios almost always represent two unrelated sites or services trying to accomplish something on behalf of users or their software. All three have to work together involving multiple approvals for the completed transaction to get authorized.

It is also helpful to remember that OAuth is about authorization in particular and not directly about authentication. Authentication is the process of a user/subject proving its ownership of a presented identity, by providing a password or some other uniquely owned or presented factor. Authorization is the process of letting a subject access resources after a successful authentication, oftentimes somewhere else. Many people think that OAuth stands for open authentication, but it’s more helpful to understand it by thinking about it as open Authorisation.

 An Early Implementer describes OAuth as similar to a car’s valet key, which can be used to allow a valet to temporarily drive and park a car, but it doesn’t allow the holder full, unlimited access like a regular key. Instead the car can only be driven a few miles, can’t access the trunk or locked glove box, and can have many other limitations. OAuth essentially allows the user, via an authentication provider that they have previously successfully authenticated with, to give another website/service a limited access authentication token for authorization to additional resources.

Additionally, OAuth 2.0 is a framework, not a protocol (like version 1.0). It would be like all the car manufacturers agreeing on how valets would automatically request, receive and use valet keys, and how those valet keys would generally look. What the valet keys could do as compared to the full function keys would be up to each car manufacturer. Just like in real life, valets and car owners don’t need to care about how it all works. They just want it all to work seamlessly as possible when they hand off the key.

How OAuth works:

Let’s assume a user has already signed into one website or service (OAuth only works using HTTPS). The user then initiates a feature/transaction that needs to access another unrelated site or service. The following happens (greatly simplified):

  1. The first website connects to the second website on behalf of the user, using OAuth, providing the user’s verified identity.

  2. The second site generates a one-time token and a one-time secret unique to the transaction and parties involved.

  3. The first site gives this token and secret to the initiating user’s client software.

  4. The client’s software presents the request token and secret to their authorization provider (which may or may not be the second site).

  5. If not already authenticated to the authorization provider, the client may be asked to authenticate. After authentication, the client is asked to approve the authorization transaction to the second website.

  6. The user approves (or their software silently approves) a particular transaction type at the first website.

  7. The user is given an approved access token (notice it’s no longer a request token).

  8. The user gives the approved access token to the first website.

  9. The first website gives the access token to the second website as proof of authentication on behalf of the user.

  10. The second website lets the first website access their site on behalf of the user.

  11. The user sees a successfully completed transaction occurring.

  12. OAuth is not the first authentication/authorization system to work this way on behalf of the end-user. In fact, many authentication systems, notably Kerberos, work similarly. What is special about OAuth is its ability to work across the web and its wide adoption. It succeeded with adoption rates where previous attempts failed (for various reasons).

OAuth vulnerabilities:

TLS can provide all those protections, but it’s up to the implementers, on all sides, to require it to be used. Coders and users should look to ensure that OAuth is running inside of TLS protection. Developers can implement code to enforce TLS use and users should be aware that TLS is being used whenever they are begin asked to input authentication credentials (just like they should anytime they are entering in credentials).

Because of the lack of inherent security binding, it’s possible for a rogue website to phish a user’s legitimate credentials during the part of the process where the user is being required to authenticate themselves to the authorization provider. For example, a user is using the first service and chooses a feature that forces an OAuth transaction to a second service.

It’s possible for the first website to fake the second website, where user authentication is often taking place. The rogue website can then collect the user’s authentication credentials and react as if the OAuth transaction had successfully taken place.

This is not just a theoretical threat. In the second quarter of 2017, a million Google Accounts were successfully phished. The defense is for users to ensure they are entering their credentials in the legitimate second website’s domain if prompted for credentials, and to avoid nebulous first websites.

There is no perfectly safe, universally accepted, SSO that works on all websites, but with OAuth, we’re getting closer.

bottom of page