The last two years has really put Open Source monitoring under the microscope. Our monitoring systems were single host based, focused on system level alerting and did not provide auto-scaling, dynamic configuration management. As we need to support continuous delivery and cloud application instrumentation we need to improve..
Starting with famous presentations saying that #monitoringsucks, the community started looking at how to bring monitoring up to date. This started the #monitoringlove hash tag on twitter complete with its own t-shirt. Jason Dixon did a good overview in 2012 on The state of Open Source Monitoring which gave a good summary as to where we were at. The birth of the Monitorama conferences spawned a face to face venue to continue the discussion and share how modern monitoring has improved and how we can now actually run automated monitoring setups that monitor everything from applications, platforms and even monitor for security vulnerabilities in near real-time. Capacity planning needs to not thought of a quarterly thing, but rather something we do on a regular basis as part of running our business.
Let's talk about monitoring. What tools do you use and how could they be better? How are tools such as Sensu, Collectd, Graphite and Elasticsearch solving problems that we never knew existed 10 years ago?
• How are you collecting metrics?
• How are you dealing with alerting and notifications?
• How are you visualizing and instrumenting your platform?
• What libraries are you using to instrument your code? Coda Hale's Metrics library? Statsd? something else?
Looking forward to hearing your thoughts and share some of our approaches.