Apache: Big Data 2016

Monitoring provides information on system performance, however tracing is necessary to understand the performance of individual requests.

Systems such as Zipkin, Dapper, and HTrace provide distributed tracing; as does CQL request tracing in Apache Cassandra. Such tracing is invaluable when diagnosing individual requests, yet knowing which database queries to trace and why they were made still requires deep technical knowledge. And while each solution provides insight, the problem of providing a single tracing view across a distributed application stack remains.

This talk will introduce using Zipkin to record Cassandra request traces, to provide a single tracing view from HTTP server to database. Starting with CQL request tracing; we will move onto Zipkin, ongoing work to record request traces via Zipkin, and the efforts of the OpenTracing community create a common tracing API.