Logging makes perfect - real-world monitoring and visualizations (slides)
Here are the slides for my BuildStuff talk in Lithuania - I hear the video of the talk will be available soon as well.
How to keep a real-time, low-latency and high-stakes system up and running and well-monitored? how to investigate failure cases as they happen? and how to even know something is wrong before it's too late? With logs of course. Lots of them. And then some cool stack to do stuff with it.
Forter is a company with a Decision-as-a-Service product that deals with many e-commerce transactions in real time and answers a simple but hard question: "is this a fraud attempt or not?". And if we were wrong, we pay.
In this talk I will show how we are using various technologies to power our service and keep it high-available and well under control. Among the technologies I will discuss are Apache Storm, Node.js, Riemann (state machines in Clojure, yay!), collectd, D3.js and of course the ELK stack (and beats!). Some integrations which will be mentioned include PagerDuty, Slack, Jenkins and GitHub.