December 2018 Summaries

Honeycomb offers a comprehensive approach to application performance monitoring (APM), metric analysis, and log management by providing a unique event-based model that enhances observability and debugging capabilities. Unlike traditional tools that aggregate data and may limit dimensions due to storage constraints, Honeycomb encourages the addition of multiple dimensions and retains access to raw data, allowing for more detailed analysis and faster problem resolution. The platform integrates seamlessly with OpenTelemetry for quick setup and supports sophisticated queries and alerts, enabling users to address complex application issues beyond just latency concerns. Honeycomb's model also improves the efficiency of data storage and querying, reducing the noise typically associated with traditional logging systems. By focusing on observability, Honeycomb empowers users to gain deeper insights into system operations, making it a valuable addition or alternative to existing APM, metric analysis, and log management tools.

In a two-part blog series, the author addresses the complexities and considerations surrounding the cost of observability in tech infrastructure, emphasizing that observability spending should typically range from 20-30% of infrastructure costs. They discuss the hidden costs associated with underinvestment, such as increased engineering time and reduced productivity due to suboptimal tool usage, and critique vendors for lack of transparency in their pricing models. The author argues that while cheaper solutions may seem attractive, they often lead to inefficiencies and increased expenses in the form of additional engineering resources. Organizations are encouraged to use effective tools to maximize engineering productivity towards core business objectives, and the importance of aligning spending with business needs and customer expectations is highlighted. The series also promises to delve into what constitutes a comprehensive observability stack in a subsequent post.

In a review of a SPOOKY HALLOWEEN meetup cohosted by LaunchDarkly, several talks on incident response strategies highlighted the importance of effective communication and prioritization during crises. Paul Biggar's discussion emphasized the necessity of assigning a leader to manage incidents and stressed transparency with customers as a key factor in retaining them post-crisis. Marc Deven illustrated the value of transparency and business insurance during a glitch at a company that mistakenly priced products at a penny, emphasizing how effective incident management can become a point of pride and influence company culture. Eric Pollmann from Google shared insights into handling a data push disaster, underscoring the need for robust revert procedures and the realization that even the best systems require hands-on problem-solving. The meetup also underscored the benefits of using gating software, like LaunchDarkly, to manage risks associated with software releases, and the collective learning from these incidents was seen as essential for improving incident response protocols at Honeycomb.