Abstract
Troubleshooting or monitoring high availability, high throughput systems has to be done carefully, adding the pervasive use of multicore technology and we are in a situation where an increasing numbers of problems can only be diagnosed via tracing. This talk will describe why certain characteristics of tracing tools are important, how to develop systems with built-in diagnostic, best practices on tracing and how to analyse the data. Industry initiatives such as the common trace format (Multicore Association, Linux CE Workgroup) and trace analysis at the Eclipse Foundation will also be described in a hope to create an industry de-facto way of developing SW with diagnostic in mind. Programmers, experienced system administrators and managers should find relevant information on how to improve troubleshooting of high throughput/availability system, embedded systems and HPC.
Biography
Manages developer tools at Ericsson where he is responsible for software engineering improvements, creation of research projects, open source initiatives and collaboration with other companies. He is behind many improvements in GDB, LTTng, Eclipse CDT, Linux tools. Dominique has been speaking at various conferences such as the MulticoreDev Conference, EclipseCon, LinuxCon, Embedded System Conference, Embedded Linux Conference, GNU tool chain summit, Open World Forum, etc.