October 2019 Summaries

Running experiments helps bridge gaps between technical and business teams by validating user engagement patterns and application behavior. Experiments can validate that development efforts align with business objectives, using metrics to test hypotheses about feature development processes. To create a good experiment, consider best practices such as creating a culture of experimentation, defining success, ensuring statistical significance, proper segmentation, recognizing biases, and conducting retrospectives. Feature flags are crucial for well-run experiments, with best practices including considering experiments during the planning phase, empowering others to run experiments while avoiding technical debt, and removing feature flags after an experiment is completed or not released.

LaunchDarkly has released an official Terraform provider, allowing development teams to manage LaunchDarkly resources using Terraform's declarative configuration language. This integration enables users to configure LaunchDarkly resources alongside their infrastructure stack and simplifies the management of feature flags, environments, projects, and more. The Terraform provider aims to streamline tasks that can be automated and standardized while still allowing users to leverage LaunchDarkly's web application for other functions.

LaunchDarkly introduces a new feature called the Experimentation add-on, which allows teams to measure the impact of their flagged features on business metrics. This add-on enables multivariate testing and helps in making data-driven decisions for rolling out the best feature variations. It supports full-stack metrics and can be used by operations, frontend, and backend teams. The Experimentation add-on is available to LaunchDarkly Pro and Enterprise plan customers and can be accessed by contacting their account team.

Trello and LaunchDarkly have integrated their features, allowing teams using Trello for project management to use LaunchDarkly's feature toggles directly within Trello cards. This integration enables teams to attach feature flags to stories or epics, making it easier to track the status of features and deploy them directly from Trello cards. The LaunchDarkly Power-Up respects environment rules and can be installed from the Trello Power-Ups directory.

Karishma Irani, Principal Product Manager at LaunchDarkly, discussed how their feature management platform enables product managers to experiment with and validate new features in a production environment during the September Test in Production Meetup. Feature flagging allows for experimentation by testing different variations of a particular feature, such as colors of a button or placement of page numbers. This helps PMs control certain aspects of feature testing and releases, leading to faster value delivery and better customer feedback. LaunchDarkly's custom roles provide granularity in controlling access to features, while its experimentation capabilities allow for rapid iteration and scaling.

At the September Test in Production Meetup held at Microsoft Reactor in San Francisco, Humio discussed how they use their own log management platform to test in production. The company's developers mitigate risk by pushing new features through multiple "levels of production" and building observability into their application. They also utilize feature flags and profiling tools to monitor real-time execution of queries and track the performance of their product. The key takeaway is that it's crucial for applications to be built with observability in mind, ensuring that developers can understand what's happening throughout the entire lifecycle and quickly remediate any issues that arise.

LaunchDarkly has introduced an integration with Datadog to help teams visualize the impact of feature deployments on their application and infrastructure metrics. This integration pushes events from LaunchDarkly and overlays them in the Datadog Events dashboard, improving visibility and context of changes across all systems. It enables easier detection and resolution of harmful changes to system performance resulting from feature flags being turned on or off. DevOps teams can now have LaunchDarkly context inside their Datadog dashboards, allowing them to understand what triggered issues. The integration is designed to improve the ability for teams to monitor and control the health of their production systems while enabling continuous delivery.

LaunchDarkly's Lunch & Learn panel discussion in Chicago highlighted the transformative power of feature management, particularly through the use of feature flags. The participants, including LaunchDarkly CTO John Kodumal and customers Bala Mukund from Charles Schwab and Brian Issleb from Seismic, shared how they're using LaunchDarkly's platform to ship software faster with less risk. They touched on themes such as targeted rollouts (Progressive Delivery), managing feature flags at scale, and rapid, "risk-free" delivery. Charles Schwab discussed its digital transformation process and the role of feature management in improving their software deployment process. Seismic explained how they tailor feature rollouts to each customer using feature flagging for coordinating releases.