Don’t rely on fragile in-house systems

Building and maintaining even a minimum viable in-house feature flagging tool comes with a lot of complexity, scaling issues, and unforeseen variable costs. As demands for more advanced features inevitably grow, those problems worsen. Some of these advanced features may include the ability to:

• Standardize efficient release processes across every platform and application (you get this with LaunchDarkly's 26+ SDKs for major languages/platforms).
• Detect and remediate application errors and performance issues before they impact customers. 
• Target specific features to select audiences in real-time (this is a big one).
• Attach metrics to every feature release through experiments to determine a winning variation. 
• Ease the burden on developers and enable non-technical teams to release features with custom roles and a simple UI.

Supporting those advanced capabilities in-house is expensive and labor-intensive.

There are also critical architectural considerations. Engineers must ask—What happens when my in-house feature flag system goes down? With LaunchDarkly, you don’t have to worry. Our architecture provides several layers of resilience across the platform to ensure your systems and clients stay operational at all times, even in the unlikely event of an issue with the LaunchDarkly service.

Building and maintaining a high-performing, reliable flag architecture that works seamlessly across all platforms is also extremely challenging. But with LaunchDarkly's global flag delivery network, you can:

• Initialize within an application in 25ms
• Enact flag changes across all platforms and environments in 200ms (i.e., release or roll back changes in 200ms)
• Enact flag changes across the world in 200ms via 100+ global points of presence

Rather than support all of these daunting requirements, engineering teams trust LaunchDarkly for feature management. And they get back to doing what matters: building products that customers love.

LaunchDarkly’s simple UI and custom roles allow product teams to work hand-in-hand with engineering teams to release new features. Without making any changes to the code, product teams can do things like:

• Ease the burden on developers and release features on their own time. 
• Test new features in production before rolling them out to the entire user base. 
• Target specific features to select audiences in real-time.
• Conduct A/B tests and multivariate experiments with high statistical rigor and intuitive results. 
• Monitor features to gain insights into feature adoption, performance, and user engagement.

At a minimum, we believe a feature management system should:

• Enable real-time flag changes, anytime, anywhere—no restart, redeploy, or waiting regardless of where your app is located.
• Let you use flags seamlessly and consistently across all your applications and platforms on the server-side and client-side.
• Offer automated fail-safe mechanisms during incidents.
• Provide team management and governance capabilities for compliant and consistent releases.

Building and supporting even these minimum requirements in-house involves considerable engineering effort. LaunchDarkly's architecture provides real-time flag changes across environments through a global flag delivery network, robust resilience to ensure clients stay operational at all times, 26+ SDKs for major languages/platforms, and a streaming architecture, all of which enables you to:

• Enact flag changes in real-time across all platforms and environments. 
• Evaluate flag rules, no matter the circumstances, through self-healing SDKs that retrieve and cache flag values and targeting rules to memory.
• Maintain application stability by instantly rolling back buggy features without a redeploy.

If your team lacks expertise in security practices, it increases the risk of security vulnerabilities in your homegrown feature management system.

Whereas LaunchDarkly ensures industry-leading security for 5,000+ customers, including the U.S. Federal Government. Our dedicated security team stays current with security threats, implements best practices to maintain a high level of security, and holds certifications like ISO 27001 and SOC 2. Additionally, LaunchDarkly offers security features such as federated authentication and policy-based authorization.

We operate our data privacy program in accordance with privacy laws including GDPR and CCPA. LaunchDarkly provides comprehensive privacy and security protections that enable our customers to use our products in compliance with HIPAA and FedRAMP.

Monitoring and appropriately retiring feature flags can be a laborious task that can pose great risk if flags are retired prematurely.  With LaunchDarkly, leverage flag status indicators to know when it is safe to clean up flags—and do it with one simple click. LaunchDarkly’s code references also help you quickly identify the repository and line of code to be updated. You can also view a running tally of all flag changes in any given environment in LaunchDarkly’s audit log.

In our view, NO.

The biggest mistake people make when deciding whether to build a feature management system is they focus too much on the cost of the initial build (fixed) while sorely underestimating the maintenance costs (variable). For many, feature management becomes critical infrastructure. Thus, the initial build is never the end. There is always ongoing work to:

• Maintain
• Troubleshoot
• Secure
• Integrate with changing tools in the stack
• Respond to user requests
• Extend capabilities to additional teams
• Etc.

When building your own system, you are, in effect, making a permanent decision to commit FTEs across several roles to managing that system. Depending on the scale and complexity of your deployment, that can be over $1M/year by itself.