An often overlooked and under-invested area of the DevOps transformation is monitoring, and specifically real user monitoring. This talk is about how taking small steps towards enabling real user monitoring can result in huge strides towards improvements for DevOps teams.

Monitoring, which is a part of the DevOps toolchain, isn’t given the right kind of treatment. It is mostly included as an afterthought and is a hodgepodge of custom telemetry, impulsively built instrumentation, and a handful of mark-and-measure scripts for the front-end.

The landscape of monitoring tools consists of infrastructure and system monitoring tools, network monitoring tools, log monitoring tools, application monitoring, performance monitoring, synthetics, and various others. These form a complex pool of software by themselves, and their output quickly overwhelms engineers and engineering teams with numbers. Enormous time and effort is spent by engineering teams in converting these numbers into intelligent and actionable insights. Only to be made extinct by the next failures in production or new wave of tools and techniques introduced to the DevOps ecosystem.

The perpetual priority for DevOps is user experience. Degradation in user experience should always be priority #1. Therefore, all monitoring, alerting, and paging systems must be tuned to focus on metrics about user experience.

This talk explores many tips, tricks, and techniques that will help you complete your transformation to a truly DevOps-ready team/organization.


Comments are closed.

Dana Luther at 18:20 on 24 Jan 2020

This was a great explanation of why tracing and real-time monitoring are both so important in modern infrastructures. It was clearly presented and very engaging.

Intresting, nice to see theses explanation about tracing and their importance. But I find it a bit too slower,i think the journey could have been shorter and some extra technical information could have bring

Koen Cornelis at 23:40 on 24 Jan 2020

Nice story & slides. Not giving five thumbs up because the abstract doesn't quite cover the talk.

Steve Winter at 08:10 on 25 Jan 2020

Well structured session, but I felt it moved too slowly - the background could have been covered more quickly which would have left time for more technical content.

You also did well not to make it a ‘vendor session’ (though the ‘sign up for a free trial’ on the bottom of every slide was a little unnecessary).

Very interesting topic and a good decision to also explain tracers.

The intro felt a bit long and slow. I think it would help to see the example on a deeper level instead :)

Very good high level explanation on end-to-end distributed tracing.