cschneid/statsd-instrument

The statsd-instrument library lets Ruby applications track performance metrics like response times, error counts, and unique user visits, then send that data to a monitoring system so you can see trends on dashboards. It was built at Shopify to monitor their production traffic. It works by sending small packets of metric data over UDP, a lightweight networking protocol that doesn't wait for confirmation that the data arrived. This means logging metrics won't slow your application down even under heavy load, but occasionally a data point might get lost in transit. That tradeoff is intentional, the end goal is spotting trends over time, not keeping perfect counts. The data flows into a tool called Graphite that automatically rolls up fine-grained numbers into broader averages the older they get, keeping storage fixed-size while still showing meaningful patterns. The library offers two ways to instrument your code. You can call its methods directly wherever you want to record something, or you can use its metaprogramming feature to attach metrics to existing methods from a single configuration file. The metaprogramming approach lets you measure how long a method takes, count how often it's called, or track success versus failure rates without cluttering your business logic with monitoring code. It supports counting events, recording instantaneous values like queue depth, tracking unique values like customer IDs, and timing how long operations take. A team running a Ruby web application at scale would use this to catch performance regressions, watch error rates climb, or spot traffic spikes before customers complain. It also ships with testing helpers so developers can assert in their test suites that the right metrics get recorded at the right moments, and it behaves sensibly across environments, actually sending data in production, logging to the console in development, and silently ignoring everything in tests.