Every organization is embracing DevOps to one degree or another. The business impact of shipping software quickly and adapting to market needs is so immense that it has become a requirement—you’re either heading toward DevOps or heading toward bankruptcy. Yet, while our need for speed has increased, so has our need for security—and combining both is far from an easy task.
The key to achieve the DevOps velocity is developer autonomy. Developers and their surrounding application teams need to make daily decisions about what to build and how to build it, and they need to do them on their own. Bottlenecks appear when teams are required to get external approval, slowing down the process and eliminating the dev team’s sense of ownership. The resulting effects are subpar outcomes in applications and attrition of top performing developers.
Yet, security practices, unfortunately, often depend on such reviews. Security audits and reviews are the bread and butter of most security organizations, requiring experts to determine if a decision is secure or not. For security to work at the speed of development, this practice has to change and give developers the power to make security decisions. The only alternative is insecure software and deceleration of software version development.
How can we achieve such developer empowerment for continuous secure software development? Let’s review some key practices, and look at ops practices for inspiration.
Mindset Shift: From Super-Hero Operators to Multiplier Engineers
Site reliability engineering (SRE) teams hold the same responsibility for a service’s uptime as the sysadmins that preceded them. The change in the role came from how they achieved this resilience. SREs don’t handle most incidents, nor do they review all code that affects reliability. Instead, they engineer the tools to empower developers to handle oncall and build operable software the right way. An SRE should only be pulled in for the extreme cases, be it an incident or architecture review.
Security teams must operate in the same manner. Their job must be tactical, working to reduce their involvement in the process and giving more responsibility to the developers. Cases will arise where their expertise is needed, but general inquiries should be kept to a minimum.
Tooling Change: From Security Tools to Developer Tools
Before DevOps kicked in, app performance monitoring (APM) was owned by IT, who used synthetic measurements from many points around the world to assess and monitor how performant an application was. These solutions were powerful, but their developer experience was horrible. They were expensive, which limited tests developers could run. They excelled in explaining the state through aggregating tests, but offered little value to a developer trying to troubleshoot a performance problem. As a result, developers rarely used them.
Then, New Relic came on the scene, introducing a different approach to APM. Their tools were free or cheap to start with, making it accessible to all dev teams. They used instrumentation to offer rich results in developer terms (call stacks, lines of code), making them better for fixing problems. This new approach revolutionized the APM industry, embedded performance monitoring into dev practices and made the web faster.
The same needs to happen for application security. If developers are the ones making impactful security decisions, then the most important users of security tools are no longer security folks—but rather developers themselves. Security tools should therefore focus on achieving developer adoption first and foremost, even when it comes at the expense of more effort by the security person involved.
Governance Change: From Blocking to Monitoring
One of my favourite DevOps adages is, “If it moves, measure it. If it doesn’t move, measure it in case it moves”. It embodies just how critical it is to successfully monitor your running application. While CI/CD pipelines help you get software to production, monitoring gives you the confidence to do it at speed. You don’t need to worry so much about shipping a bug when you’re confident you’ll find out about it quickly and revert or roll out a solution.
Security needs to embrace this approach. Finding out a security vulnerability has just been deployed and acting on it is almost as good as blocking it just before it was deployed—and far less disruptive. But to embrace this, you’ll need to invest in monitoring.
Developers are the first step in the application process. It makes sense that they are also the engineers of security solutions. Empowering developers to make security-first decisions allows IT teams and security professionals to address different parts of the business, saves time in the development process and ultimately creates stronger applications and outcomes.
