FlightAware is the world's largest flight tracking and data platform; they fuse hundreds of global data sources to produce an accurate and consistent view of flights worldwide. They make this data available to travelers, business aviation providers, and airlines through web and mobile applications and different APIs.
The team tested several monitoring setups to track, visualize, and alert on the quality of flight predictions. First, they considered using Zabbix because it was widely used at FlightAware. They quickly realized that Zabbix could not handle the load of all metrics they wanted to track. After ruling out Zabbix, they started looking at Grafana as a visualization and alerting tool and Prometheus as their database backend. Unfortunately, Prometheus did not support JOINs and could not store all the data they wanted to collect.
Timescale's excellent performance, impressive documentation, Grafana support, and the fact that FlightAware already used PostgreSQL in many applications were the main reasons they decided to add Timescale to their tech stack. They also took advantage of Timescale's continuous aggregates feature to speed up their queries significantly.
Continuous aggregates are what well and truly sold me on Timescale. We went from 6.4 s to execute a query to 30 ms. Yes, milliseconds.
Caroline Rodewig, Senior Software Engineer at FlightAware