Monitoring at Zalando

This month, we'll have Jan Mussler from Zalando speaking about monitoring. In his own words:

In late 2013, Zalando’s technology team maxed out on our Icinga/Nagios infrastructure both performance-wise but (especially) in terms of manageability Taking advantage of our annual Hackweek—a weeklong event for Zalando technologists to build, play and experiment independently—a group of us engineers built ZMON: an open source monitoring solution to provide teams with a performant, reliable, and flexible tool for monitoring all levels of our platform. ZMON is equipped to monitor low-level system metrics via SNMP/NRPE and HTTP requests to exposed metrics, as well as higher-level KPIs via SQL and much more using Python expressions as tasks. Its base components include a scheduler, Redis for state and queue, and a distributed set of workers responsible for evaluating checks and alerts. On top of these are a frontend component for Dashboards and Alerting (it includes Grafana to make the most of your time series data) and KairosDB as a metric store. We made trying out ZMON as easy as spinning up a Vagrant box :-)

This talk will cover in more detail how ZMON works and what we learned building it. I will dive into how our tech team uses it in production to monitor our Zalando platform, which spans across two data centers with hundreds of deployed application instances and databases of various kinds. With an engineering department surpassing 700 engineers and non engineers we will also look into the organization of checks and alerts and how engineering teams and incident teams benefit from our alert model, including inheritance, and reuse of checks.

About Jan

Jan Mussler joined Zalando’s database engineering team in 2011 after graduating RWTH Aachen University. He has worked on monitoring and improving PostgreSQL performance and availability, and on Zalando's EventLog infrastructure using both SQL and NoSQL storage options. Since we began developing our own monitoring solution called ZMON in the end of 2013, it became the main focus and remains until today.

Meeting space, beers and pizza generously provided by Zalando.

Please note - There is no access to Zalando before 19:00. Talks will start at 19:15