This site is the archived OWASP Foundation Wiki and is no longer accepting Account Requests.
To view the new OWASP Foundation website, please visit https://owasp.org
Difference between revisions of "AppSensor Cheat Sheet"
| Line 17: | Line 17: | ||
= Purpose = | = Purpose = | ||
| + | |||
| + | Attackers take a finite number of attempts to find vulnerabilities, and additional attempts to craft an exploit. This may be scores or hundreds of attempts; or tens or hundreds of thousands if using automated tools. If you can detect some of these steps, you identify an attack and take some immediate action. | ||
Distribute detection points in the application to detect very suspicious and blatantly malicious use | Distribute detection points in the application to detect very suspicious and blatantly malicious use | ||
| Line 25: | Line 27: | ||
= Pre-requisities = | = Pre-requisities = | ||
| + | |||
| + | * application that is fully tested (e.g. functional, stress) since the AppSensor concet relies on detecting abnormal behavior, and errors in an application could be misinterpredted | ||
| + | * secure application (using a secure development life cycle) to minimize the existence of unknown weaknesses | ||
| Line 32: | Line 37: | ||
It is very important not to make detection overly complex, and vital not to include events which could be triggered by normal, even accidental, behavior. | It is very important not to make detection overly complex, and vital not to include events which could be triggered by normal, even accidental, behavior. | ||
| − | The most commonly implemented detection points are: | + | The most commonly implemented detection points are exceptions relating to: |
| − | * | + | * |
| − | * | + | * access control (e.g. a user attempting to access a resource they do not have permission to) |
| − | * | + | * whitelist input validation (e.g. non-numerical characters for a non-editable parameter value that is expected to be a positive non-zero integer) |
= How to respond = | = How to respond = | ||
| + | |||
| + | Legally. Policy. | ||
Thresholds: | Thresholds: | ||
| Line 74: | Line 81: | ||
== No code available == | == No code available == | ||
| + | |||
| + | Watch out for | ||
| + | |||
Revision as of 20:09, 24 May 2012
Introduction
All users make mistakes, but attackers also perform actions that the user interface doesn't offer or are impossible to do without using tools like intercepting proxies. The AppSensor concept describes how to make applications attack-aware.
AppSensor will:
- Identify whether your application is under attack in real time
- Optionally initiate an immediate response action to protect the application and its data
AppSensor does not:
- Implement error handling for you (although detection points are often inherently linked with error handling code)
- Perform application security logging (but logging mechanisms could be used to record detected events)
- Protect you if you have existing vulnerabilities (fix known vulnerabilities and implement secure-SDLC activities to avoid new ones)
Where possible implement the recommendations in the other cheat sheets as thoroughly as possible first, before considering the ideas within AppSensor. Robust error handling mechanisms and security event logging should already be part of well designed quality software.
Purpose
Attackers take a finite number of attempts to find vulnerabilities, and additional attempts to craft an exploit. This may be scores or hundreds of attempts; or tens or hundreds of thousands if using automated tools. If you can detect some of these steps, you identify an attack and take some immediate action.
Distribute detection points in the application to detect very suspicious and blatantly malicious use
Collect these event event data
Define a very low tolerance threshold, and take a predefined automated action once a user exceeds the limits of reasonable use
Pre-requisities
- application that is fully tested (e.g. functional, stress) since the AppSensor concet relies on detecting abnormal behavior, and errors in an application could be misinterpredted
- secure application (using a secure development life cycle) to minimize the existence of unknown weaknesses
What to detect
It is very important not to make detection overly complex, and vital not to include events which could be triggered by normal, even accidental, behavior.
The most commonly implemented detection points are exceptions relating to:
- access control (e.g. a user attempting to access a resource they do not have permission to)
- whitelist input validation (e.g. non-numerical characters for a non-editable parameter value that is expected to be a positive non-zero integer)
How to respond
Legally. Policy.
Thresholds:
Typical responses, that the application may already support in some manner, are:
- Change monitoring of the user (e.g. increase logging level)
- Raise an alert
- Add time delays
- Log a user out (and possibly lock the account)
More advanced responses could include
- Changing a function (adding a CAPTCHA, ??? )
- Disabling a function (for the user, for a group of users, for all users)
- Affecting behaviour of another systems (e.g. goods despatch held, firewall blocks IP address)
- Altering user properties (changing their credit level)
How to
Software acquisition
In your own code
No code available
Watch out for
Related articles
Other Year of Security for Java Week 18 - Perform Application Layer Intrusion Detection
Authors and primary contributors
Colin Watson - colin.watson[at]owasp.org
OWASP Cheat Sheets Project Homepage
