For the large-scale projects we get involved in, security is often a major consideration. When taking over an existing concern, one of the first areas we audit is application security. More often than not there are holes. Usually these are simple oversights, and sometimes the exploits are only available to technically savvy people with a lot of time to spend on an attack. However the most common flaws we see in large applications are caused by mistaken assumptions. The most dangerous and most common of these is the assumption that an authenticated user is an authorised user.
In simple terms, authentication is the process of identifying yourself and proving it. Online, this is usually achieved with a combination of a username and password. This can be considered equivalent to entering a building with a keycard. Assuming your identification doesn’t get stolen, you can securely verify who you are. This is authentication.
Lets assume the building you’ve just let yourself into is a large, shared office block. You now have free roam of the circulation spaces in the building, but you wouldn’t expect your keycard to get you into other peoples offices. Why not? The keycard is still doing the same job, it authenticates that you are you. It doesn’t and shouldn’t authorise you to do things you shouldn’t. This is authorisation, and where a lot of web developers have a large and dangerous blind spot.
Online, authentication is often handled at the application level. When you request a page or try to make a change, the application authenticates you and passes you through to the requested functionality. From there it is the developers job to ensure you are authorised to perform the requested action. There are many different patterns and paradigms around authorisation, but by far the most common we see is… none. Even with paid-for user accounts, it may be worthwhile for a hacker to sign-up to an exploitable system in order to mine user data or attack the service. In this age of Free trials, free tiers and social integration, it is critical to ensure your application developer has carefully considered authorisation and fully understands the ramifications of a missing authorisation layer.
So what’s at stake? Here’s an easy example. A user has a profile, and an edit-profile page. That page submits a modify request to your application. Inside the request is a user-id to identify which user the change is for. Without a suitable authentication layer, a malicious user can change the user-id in the form and submit changes to another users profile. One module of the application checks they are a valid user and authenticates them. A second module is then invoked which updates the profile. Without reliable confirmation that the profile being updated belongs to the user who submitted it, any user can change any other users details. Simple attacks like this are incredibly easy, and can be just as easily stopped by a solid authorisation system.