DeveloperWeek 2020

Distributed tracing has gained much popularity over the past decade, earning a reputation for being valuable in visualizing end-to-end transactions and identifying latency bottlenecks. However, tracing has much more potential beyond debugging specific issues and enabling data presentation at the granularity of a single transaction. Data from traces – when aggregated – can reveal much more, and can do so with greater precision and certainty.

This talk begins by debugging a real-world issue using distributed tracing, and expands into the profound insights trace aggregates help unlock, including bottlenecks in the entire system, sources of errors, and latency anomalies in the context of infrastructure. We also highlight the challenges in sampling traces intelligently, so systematic conclusions can be drawn from a subset of representative traces. The workshop ends with an interactive demo where the audience explores different types of aggregate trace analysis tools and challenges themselves to debug a simulated outage using distributed tracing!