Source Instrumentation for Monitoring C++ in Production
It is essential to discuss how modern C++ code can be effectively instrumented, in order to effectively monitor it after deployment. This talk will focus on portable source instrumentation techniques such as logging, tracing and metrics. Straightforward, but well designed code additions can drastically ease the troubleshooting of functional issues, and identification of performance bottlenecks, in production.
Of course when dealing with C++ performance is often critical, and so minimizing the cost of any instrumentation is also critical. Key to this is understanding the trade-off between the detail of information collected, and the overheads of exposing that information. It is also important to understand how best to benefit from advances in contemporary monitoring infrastructure, popularised by cloud environments.
This talk will open with some brief motivation towards monitoring and instrumentation. It will then walk through some practical code examples using some generic instrumentation primitives, based on proven principles employed in demanding production software.