The tools, languages, platforms, and methods used to build applications have changed drastically over the past decade. Application security practices have to change with them; otherwise, security professionals will be playing constant catch-up with attackers and cybercriminals.
What is the OWASP API Top 10?
The increase of microservices and application programming interfaces (APIs) has given rise to a new set of threats against applications. The OWASP API Top 10 documents the risks associated with API development.
Here are the vulnerabilities highlighted in the most recent OWASP API Top 10:
- Broken Object Level Authorization (BOLA)
- Broken User Authentication
- Excessive Data Exposure
- Lack of Resources and Rate Limiting
- Broken Function Level Authorization
- Mass Assignment
- Security Misconfiguration
- Improper Assets Management
- Insufficient Logging and Monitoring
Why is the OWASP API Top 10 Necessary?
Large monoliths are giving way to small, nimble microservices. Kubernetes, containers, cloud-native architectures, and API gateways are the new hotness. This shift has also led to new threats.
Let’s take a look at the differences between traditional web development and API development.
Authentication and Authorization Look Completely Different
Three out of ten risks in the API Top 10, including the first two, deal with authentication and authorization issues. These critical security functions look differently with APIs than they do with more traditional web applications.
In a typical web app, the user logs in and the server grants a session cookie. The browser stores the cookie and passes it to the server with every request, removing the need to re-authenticate. All of the user’s permissions are tied to the session ID.
APIs are made of several mini web apps all deployed on separate servers, each with its own data store. Creating a session in one server doesn’t matter much. The server receiving the next request won’t know anything about the session stored at the first server.
This complexity gave rise to the OAuth and OpenID Connect specifications. These specs outline authentication (Open ID) and authorization (OAuth) for distributed systems. Using these techniques correctly helps to keep microservices and APIs safe.
But even though authentication and authorization have changed for APIs, another aspect of them introduces a new big player in API risks: business logic errors.
APIs Talk Too Much
Traditional web applications expose one endpoint, and one executable running on the server contains all business logic. An attacker will have a difficult time mapping out all of the internal business logic of an application. With so much trial and error, it’s easier to go after vulnerabilities based on inputs, such as XSS or SQL Injection.
On the other hand, breaking an application into microservices, each with its own API endpoint, by default exposes much of the business logic to the outside world. You can watch your browser’s connections to see the individual pieces of the application as a whole. And most of the time, the public interface reveals the essential parameters and unique identifiers of sensitive data objects.
BOLA is an excellent example of this phenomenon. BOLA occurs when an attacker changes the resource ID passed into an API to retrieve someone else’s records.
Imagine a medical application that holds sensitive documents. You can view your documents in PDF form. Since APIs are used to serve up your data, the client app passes an ID through the URL. You accidentally (because we know you’d never do this on purpose) transpose the numbers in the URL. All of a sudden, another person’s medical information stares back at you. You’ve discovered a BOLA vulnerability.
Another example of APIs talking too much is with claims-based authentication. Typically, JSON web tokens (JWT, pronounced “jawt”) are used to send the current user’s permissions, referred to as “claims.” These tokens are easily decoded and updated.
Imagine receiving a JWT using OAuth and decoding it. You notice that one of the attributes on the token is “admin: false.” Now you know how the application determines your permissions. You change the attribute to “admin: true” and send it back to the application with the next request. If the API doesn’t follow the correct token signing and verification procedures, you’ve now become an administrator and can gain access to all sorts of goodies.
Make sure your APIs aren’t leaking too much information. Be aware of what your API’s clients see and determine the risks and mitigation of each piece of data.
“Loose lips sink ships” was a famous security warning during World War II. In today’s world, loose data sinks APIs.
Microservices and APIs Are (Too) Easy to Create and Deploy
New development methodologies come with cultural changes within organizations. Developer autonomy and delivery speed are emphasized, leading to many development teams deciding how to implement APIs and when and where to deploy them.
These methods lead to greater flexibility, agility, and scalability but could also create API proliferation and increased risk. Number nine on the API Top 10 is Improper Assets Management. Not knowing where all of your APIs are and what they do can lead to serious application vulnerabilities.
For example, in creating and deploying a new version of an API, an endpoint running the old version could still exist. Should the old version contain a security vulnerability, attackers could exploit it without your knowledge. Building an inventory using API discovery techniques allows organizations to find these endpoints and shut them down properly.
Businesses need to be agile and responsive. But they must also be aware of the side effects these strategies create and design processes to keep things under control.
If You’re Building APIs, Don’t Forget the API Top 10
Building microservices and APIs is distinct from building traditional web applications. Differences in environments, frameworks, and development techniques solve some problems while introducing new ones.
Get familiar with the OWASP API Top 10 and use it to check your applications. Add it to the regular security checkpoints you already have. Then you’ll be confident your APIs are safe and secure.
Traceable Defense AI helps protect against the OWASP API Top 10. View a recorded demo to see it in action.
About the Author
Justin Boyer is a technology writer and regular contributor to The Inside Trace.