Berlin 2013 - Proposal

Gold sponsors

Back to proposals overview - program

We love alerts at 2am


When I joined Prezi I was promised that I don’t have to be on-call ever. We promised this to everyone who joined our team. Last year we decided that we’d like to be on-call and we are really happy about this.

Through the story of the evolution of our monitoring system I’d like talk about culture.

Last year we doubled our active user count. As more users mean more responsibility, we had to concentrate on improving availability while simultaneously scaling our backend. One important component of this process is improvements in our monitoring system. I’d like to show how a good monitoring system enables happy and effective component owners.

Originally we had the traditional, entirely separate Dev and Ops silos: we contracted with a company that was responsible for operations, and they used their own monitoring system. We added our own application metrics to this system, and when they received an alert they tried to solve the issue. As you can imagine, this arrangement had some problems. We have since got rid of these silos, bringing our ops in house, and are now much happier.

In this talk I'd like to present what we did to improve our monitoring situation in small steps: we created our own monitoring system that makes it trivial for developers to add checks of our complex system. We introduced the concept of component ownership, and even though our developers must be woken up for some alerts at night, everyone is more satisfied than before.

The talk will go into detail about the system itself as well as the social aspects of the arrangement.


Gabor Veszi

blog comments powered by Disqus
Zeroturnaround HP

Gold Sponsors

Gutefrage Immobilien Scout The unbelievable Machine Company Nokia Here Engine Yard Idealo Netways Axel Springer

Silver sponsors

InnoQ FriendScout 24 Serena Cassini Leanovate IT-Agile Cloudbau Stanford Ignite Paris Tarent OpsWorks E-POST

Evening sponsors

SysEleven github