Secure Software Begins in the Development Process
Though there are many risk mitigation options for the software management lifecycle (firewalls, intrusion detection/prevention systems, etc.) these dont address the root cause of the problema software development lifecycle (SDLC) that is not integrating security at each phase.
Perimeter defenses are simply unable to stop most software attacks, because abusers are focusing on the application layer; shying away from attacks against the system and networking layer. Secure applications are a software development challenge that will never be solved until security is addressed as part of the software development process.
Only a concerted effort by the software development team to produce more secure applications will protect you from exploitation.
A New Business Twist
A successful application used to be defined as an application that solved the problems it was designed to solve, and did so with some reasonable performance. From a business standpoint, however, this definition must be rethought. Not because performance alone is a poor objective but because applications can be leveraged to exploit and destroy your business modelthe very model youre building it to assistand this must be considered during development.
Though applications can be retrofitted for security post-deployment, this is a very expensive and ineffective approach. All software development methodologies (i.e., iterative, agile waterfall, etc.) will benefit from a focus on security practices. In each case, incorporating security-based activities in each phase of development will improve quality and resistance to attack in the final product.
Here are the main activities you should incorporate in each phase:
Defining requirements explicitly, including how applications should and should not interact with their environment, ensures that projects start with the right foundation.
When constructing security requirements consider what a system must not do. Think abuse case rather than use case. For example, describe how a malicious user might attack the system or misappropriate the data the application touches. This will force design decisions like when and where to use encryption to protect your data.
When defining security requirements, you also need to understand risk. In particular, the business risk of a successful exploit against the application and how that exploit may affect users. What business processes would be compromised, damaged, or controlled? The costs of liability, patch development, and damage to brand and market share need to understood and considered in this phase.
The most expensive security defects are those introduced during design. Design considerations include both architectural issues as well as individual component design.
At the system level, the objective is to use components and configurations that reduce the applications attack surface. Threat modeling is a very useful activity during this phase as it allows you to identify potential threat quickly and create persistent models with which you can test various configurations and known risks quickly with different architectures.
The best tactic here: Education. Train your architects and developers on proper use of libraries and patterns, showing them how to implement each component securely.
Clear requirements and well-tested designs will make for a less troubling and vague development process. But even with those great starts, developers are prone to mistakes like all humans; especially when they arent properly trained.
Educate your team on proper error handling routines by seeking secure coding training. Such courses can show developers how to avoid dangerous code constructs, and how to properly validate input, use encryption and create secure data transport mechanisms.
Also, capture knowledge during development. One of the largest costs of secure coding is the cost associated with making the same mistake over and over again. Conducting security code reviews is a very valuable activity and one of the best places to find that knowledge that you should capture in an internal knowledge base or defect management system.
Testers need to think differently and learn new activities to effectively find security defects. Testers need to give more attention to the applications environment, network connections, configuration, customization options, and non-functional operation (pushing the application to do things it was not designed to do, for example).
Instead of thinking Does the application do 'X', as its supposed to? testers need to think What if the application did 'Y'? and then test for that. It is in the unintended behavior of applications where security defects hide.
Deploy and Maintain
Secure maintenance practices emphasize the importance of understanding the existing security infrastructure and what risk mitigations are already in place. Maintain documentation and utilize monitoring tools to track any changes and audit abnormal behavior patterns for risks they impose on the overall security of the system.
These practices become critically important when a security bug is discovered and must be fixed and patched in an already deployed application. While fixing the defect is paramount, the use of a well-crafted incident response plan will help ensure a smooth process that fixes the problem with minimum risk of introducing additional security defects in the application.
This will also minimize user down time and make for happier clients. Often a high quality security response program is just as valuable as inherent security quality in your application.
Ed Adams is the CEO of Security Innovation.