September 2016 Summaries

LaunchDarkly introduces new management features for its customers who use feature flags to manage risks, releases, and software deployments. The company observed that managing feature flags at scale was difficult with existing tools. To improve this, they have introduced flag maintainer, flag tagging, flag descriptions, and rich flag variations. These features complement their existing management tools like flag statuses, custom roles, and the flag dashboard. The new features are designed to help teams manage feature flags more effectively, reduce technical debt, and communicate the purpose of each flag to team members.

LaunchDarkly has introduced Prerequisites (beta) to allow customers to control feature dependencies by evaluating Flag B only if Flag A is "on" for a particular user. This functionality enables smarter, conditional feature releases and can be managed in the feature flag's Targeting tab. Prerequisites are useful for mutually exclusive experiments, allowing users to compartmentalize experimentation without hardcoding logic into their codebase. The use of prerequisites also decouples experiments from the codebase, making it easier to modify and test features.

LaunchDarkly introduces custom targeting rules, allowing users to create conditional rules using a user attribute, operator, and value for granular feature flagging. This enables powerful and precise user targeting, such as rolling out features based on user attributes like registration date, location, or spending habits. Custom targeting rules can be constructed with multiple conditions, and percentage rollouts can also be applied to specific user groups. The new feature opens up many possibilities for advanced user segmentation and targeted feature releases.

LaunchDarkly has introduced support for multivariate feature flags, allowing users to define two or more custom variations such as strings, numbers, JSON objects, or JSON arrays. This update enables more powerful ways to manage user plans, app configurations, and multiple variations of a feature. Multivariate feature flags can be used for percentage rollouts, individual user targeting, and creating custom rules based on user attributes.

Microservices is a practice that breaks down monolithic releases into discrete services, allowing for independent release schedules. This approach was popularized by Martin Fowler and implemented by companies like Amazon and Netflix. However, microservices can make releases more complex due to the need for extensive testing between components. Feature flags offer a solution to this problem, enabling engineering teams to control various microservices effectively. By wrapping microservices with feature flags, traffic can be gradually directed to new versions while ensuring functionality in other services. This approach allows for "microdeployments" of microservices and helps unlock the value of decoupling.