Kernel Tracing in the Cluster

This proposal has been rejected.


One Line Summary

An overview of tracing services running in the cluster.


Debugging services in the cluster is a difficult problem. System-wide effects and inter-process interactions are subtle and require a large toolkit to find and diagnose. Kernel tracing is an important part of this toolkit. However, tracing in the cluster brings different requirements than tracing a single system or single process, and requires analyzing more trace data than looking for a bug in a single process or within one part of the kernel.

This talk will look at the major issues central to tracing large systems for the purposes of debugging or understanding performance. It will give an overview of requirements for a cluster-deployed system for collecting kernel trace data, and include an analysis of the major factors affecting the usefulness of the collected trace data.




  • David Sharp



    David is a member of the kernel team at Google. He is currently the Tech Lead of Kernel Tracing.