Jack Neely is a DevOps observability architect at Palo Alto Networks and has a few interesting ways of extracting value from o11y data.
We crammed into just under 25 minutes ideas like these 7 takeaways:
Reasserting the Need to Monitor Four Golden Signals: Focus on latency, traffic, errors, and saturation for effective system monitoring and management.
Prioritize Customer Health: in Jack’s words, the 5th golden signal. Go beyond traditional metrics to monitor the health of your customers for a more comprehensive view of your system's impact.
Apply Mathematical Techniques: Incorporate advanced mathematical concepts, like the Nyquist Shannon law and T Digest algorithm, to enhance data accuracy and observability metrics.
Build Accurate Percentiles: Implement techniques to accurately reproduce percentiles from raw data to ensure reliable performance metrics.
Manage High Cardinality Data: Develop strategies to handle high cardinality data without overwhelming your resources, ensuring you extract valuable insights.
Standardize Log Records: Use readily available frameworks to emit standardized log records makes data easier to process and visualize.
Handle High-Velocity Data Efficiently: Develop methods for collecting and processing high-velocity data without incurring prohibitive costs.
Watch Jack’s Monitorama talk here:
#50 Making Better Sense of Observability Data