September 2022 Summaries

Container orchestration platforms automate various concerns around running containers, simplifying dependencies and tasks like networking/DNS configurations, resource allocation, replica management, and workload rollout processes. Kubernetes is the most widely adopted container orchestration platform in the cloud-native world, but others include Apache Mesos and docker-compose. These platforms help manage the lifecycle of containers and ensure repetitive tasks are handled intelligently by the system. Container orchestration becomes more valuable as teams adopt a microservices architecture approach to building their application platforms. Key takeaways include automation's role in deploying and compiling applications, the adoption of Kubernetes as the most widely used container orchestration platform, and the benefits of using managed vs. self-built (homegrown) container orchestration platforms.

LaunchDarkly has introduced a new feature allowing users to create maintenance windows for flags that need temporary toggling off. This eliminates the need for multiple scheduled flag changes and is powered by Workflows, which can toggle flags on and off during specified time periods. The update aims to help teams focus more on value-adding tasks and less on unrelated maintenance work. To use this feature, navigate to a flag's "Workflows" tab, click "Create workflows," and choose "Maintenance window." Workflows are available in the Enterprise plan, with feedback welcome at [email protected].

Feature toggles, also known as feature flags, have evolved from simple on/off switches for features into powerful tools for managing the entire lifecycle of a feature in software development. Originally used to separate feature rollout from code deployment, they now enable developers to control and monitor feature releases, conduct A/B testing, perform canary releases, and manage technical debt associated with flags. As feature flagging becomes more advanced, it is increasingly important for companies to adopt robust feature flag management solutions to keep track of their flags and configurations effectively at scale.

Blue-green deployments are an application release strategy that minimizes downtime during updates by creating two identical instances of a production app behind a load balancer. One environment (blue) is live and receives user traffic, while the other (green) receives constant updates from the continuous integration server. Once the green environment is stable, production traffic can be transferred over to it. This process allows for safe rollbacks if needed. Benefits of blue-green deployments include safety in releases, increased release frequency, reduced downtime, and empowered teams. Drawbacks include complexity in infrastructure and deployments, as well as potential cost increases. Feature flags can complement blue-green deployments by providing feature-level safety during rollouts.

Feature flagging is a technique used in computer science to control the release and testing of features in software applications. Initially, it may seem simple to implement feature flags by setting boolean variables directly in code. However, this approach quickly becomes problematic as it requires redeploying the code each time the flag is flipped and makes it difficult to manage multiple flags across different environments. A better solution involves using external configuration files or databases to store flag values, which can be easily managed and updated without redeploying the code. This approach also allows for more complex flag types such as strings, numbers, and JSON objects, enabling developers to pass full configurations or content changes through flags. As feature flags become more powerful, managing them can become a full-time job. Feature management platforms like LaunchDarkly offer additional capabilities beyond simple feature toggles, including multivariate flags, progressive rollouts, attribute targeting, and A/B testing. These features enable teams to test the impact of changes, manage access to beta features, and align releases with company goals while improving productivity and reducing risk in software development.

The article discusses the shift from using statistical significance (stat sig) in online experimentation to alternative approaches, such as expected utility. Statistical significance helps reduce false positives but may miss valuable gains due to increased false negatives. Expected utility focuses on making the best decision for each experiment based on available evidence. The article presents a comparison between stat sig and expected utility using an A/B test example and demonstrates that expected utility generally results in more correct decisions. It also addresses Type I and Type II errors, showing how expected utility handles them differently than statistical significance.

In a recent webinar featuring Gene Kim, topics such as metrics to measure, the role of architecture in software, and company culture were discussed. The Q&A session touched on various subjects including writing a book versus developing software, inter-team dependencies, and how high-performing teams resemble well-constructed software functions.

LaunchDarkly, a feature management platform, held its first roadshow stop in San Francisco on June 1st, focusing on scaling software delivery with feature management. The company's Head of Product, Karishma Irani, was joined by Max Spaulding from Life360 and Taylor Stooke from Nestlé Purina to share their experiences using LaunchDarkly. Both Taylor and Max praised the platform for enabling faster deployment cycles, making releases safe and boring, and allowing separate environments for developers and QA teams. They also highlighted features like canary releases and progressive rollouts as being particularly impactful for their teams. The event covered topics such as workflows, automation capabilities, feature tech debt, team collaboration, and experimenting for business impact.

LaunchDarkly has released several key updates over the last few months, including improvements to its experimentation platform and the addition of new SDKs for Vue.js applications. The company also expanded its commitment to healthcare customers with HIPAA compliance and introduced Workflow Templates for scaling across multiple flags, environments, and projects. Additionally, LaunchDarkly added approvals by team feature and integrations with Zendesk and Zapier. Finally, the platform now allows users to sync groups from Okta to teams within their account for consistency across tools.

Knock, a company that provides simple APIs and a dashboard for introducing thoughtful notifications into products, uses LaunchDarkly to power feature flags. They recently adopted a circuit breaker pattern built around feature flags to make their services more reliable when things fail. The problem they faced was with AWS Kinesis consumers failing in ways that do not gracefully recover when they crash. To address this issue, Knock developed an Elixir supervisor called Circuit Breaker that takes inspiration from the typical circuit breaker pattern and applies it to supervision trees. This approach gives them a lot of flexibility in dealing with failures and extra load in their system. By turning off consumers strategically, they can intentionally degrade their service in order to maintain uptime on more critical parts of their infrastructure. The Circuit Breaker monitors the feature flags for each process and adds or removes child specs based on the flag status. They use Elixir/Erlang Telemetry module extensively at Knock to monitor runtime performance, including crash metrics and log messages. Future improvements include making enhancements to their Kinesis consumer library, abstracting their dependency on LaunchDarkly, and potentially releasing this tool as open source.

LaunchDarkly's Employee Resource Groups (ERGs) are essential in fostering supportive, inclusive environments for employees with marginalized identities and aligning with company values. Led by volunteers, these groups gather based on shared characteristics or experiences not typically centered in corporate spaces. The company has ERGs for various identities and experiences such as Black and African Diaspora employees, caregivers, and queer and gender-expansive individuals. LaunchDarkly rewards its ERG Leaders with additional equity grants since August 2022, following a preference for this reward type expressed by the majority of their ERG Leaders in an internal survey. The company believes that recognizing employees' extra contributions to creating an inclusive culture should be a minimum expectation for any organization.

Feature flag-driven development can increase the speed of delivering value to customers, but it also adds complexity to testing. The article discusses five types of tests that might be part of a testing strategy: unit tests, integration tests, end-to-end tests, QA testing, and user acceptance testing (UAT). Unit tests should remain unchanged when using feature flags. However, integration and end-to-end tests become more complex due to the potential variations in feature flag states. The article suggests several strategies for automated tests, such as testing the current production state and creating testing personas. Additionally, it emphasizes the importance of testing in production, allowing QA and UAT teams to validate functionality in a real environment before exposing it to all users.