September 2021 Summaries

Tech industry is known for its lack of diversity, with white men dominating the field and barriers to entry for minorities. The LaunchDarkly Foundation aims to address this issue by engaging with diverse communities and supporting organizations that promote equal opportunities in tech education. Some notable organizations include CodeNation, Girls Who Code, Ada Developers Academy, The Last Mile, Next Chapter, Vets In Tech, Operation Code, Code2040, Codebar, Code First Girls, Coders of Colour, Coding Black Females, and Code Your Future. These organizations provide free education programs for underrepresented individuals to help them enter the tech industry or programming careers.

Rick Riensche, a senior software engineer at Autodesk, shares his experience migrating from a homegrown feature management solution called Flipper to LaunchDarkly following the acquisition of PlanGrid by Autodesk. The migration involved introducing data adapter shims for write and read operations, running in dual write mode initially, and then moving to exclusive use of LaunchDarkly. The team faced challenges such as rate limiting, misconceptions about flag evaluation methods, and issues with anonymous keys. Despite these hurdles, the migration was successful, resulting in a more functional and performant system.

LaunchDarkly has introduced new enhancements in its Fall '21 Product Release, aimed at improving the efficiency and effectiveness of feature flags and experiments. These updates include a major update to experimentation add-on for defining target audience, allocating traffic to desired variations, and ensuring reliable results. Additionally, multi-armed bandits have been introduced to capture more real-time value during experiments by optimizing specific metrics like increased revenue. The platform also offers secure client-side experiments with no performance penalty or content flash through Flag Delivery @Edge. Furthermore, LaunchDarkly customers can now sync audiences of any size from Amplitude for seamless user experiences across platforms.

The article discusses the common issue of accidentally pushing faulty code to production and how to handle such situations. It emphasizes the importance of using diff commands to identify changes in the code, acknowledging that issues may not be solely due to the code but also the environment or infrastructure. In case of a problematic deployment, it advises contacting DevOps for assistance. The article highlights the inevitability of production errors and suggests ways to prevent them, such as setting up rehearsal environments similar to production, ensuring access management policies are in place, promoting an open culture where mistakes can be discussed without blame, and conducting thorough testing before deployment. It also recommends using feature flags for controlled rollouts in production.

Rebranding is a challenging process due to the subjective nature of brands and people's opinions, but it can be necessary for companies as they mature and evolve. The reasons for rebranding include reflecting changes in ideal customer profile, go-to-market strategy, pricing model, overall messaging, and creating an emotional connection with B2B buyers. Rebranding also serves as a celebration of maturity and accomplishment for internal stakeholders, such as employees and investors, and can be an important recruiting tool. Sometimes rebranding is simply needed to update the brand's image and align it with the company's current state. Ultimately, people make a brand work through their use, love, and representation of it.

Modern software development involves iterations guided by product managers, developers, designers, and data scientists who observe user behaviors and collect feedback. However, even well-tested code can encounter bugs in production that require rollbacks or hotfixes. Feature flags are a powerful strategy for quickly identifying what works and what doesn't. They allow software to be released to a limited subset of users, enabling A/B testing and gradual feature releases. This technique helps uncover potential issues before they affect all users and allows for controlled changes without fear of breaking anything.

The exponential growth of microservices has led to a complex ecosystem that requires effective platform engineering and management strategies. As more teams adopt self-serve platforms, the needs for varying workflows in continuous integration (CI), continuous delivery (CD) and deployment increase. Managing microservices can be challenging due to their interdependency, latency issues, and potential misconfigurations. To create a successful approach, it is crucial to focus on scalability, maintainability, log aggregation, monitoring strategy, and effective alert management. Feature flags can help engineering teams control growing microservices by isolating issues when needed.

LaunchDarkly, a feature management platform, uses its own software product across multiple departments. In this talk from their user conference Galaxy, Lexi Ross, Engineering Manager at LaunchDarkly, explains how they leverage the platform for API rate limiting, code migration, support ticket distribution, and more. The company's use cases may inspire fresh ideas about how to gain more leverage from LaunchDarkly. Some of these unconventional ways include using feature flags for managing API rate limits, executing major code migrations, handling massive database migrations, targeting by SDK version, granting automated trials, and distributing support tickets.