Cloudflare's network provides an enormous array of services to its customers, collecting and delivering associated data in the form of event logs and aggregated analytics. The company's data pipeline is designed to handle up to 706M events per second, representing a 100x growth since 2018. To efficiently process this high volume of data, Cloudflare employs various techniques, including downsampling, which involves reducing the amount of data while retaining valuable information. Downsampling allows for the retention of information from more than half a billion events per second, even when some data must be dropped. The pipeline uses a combination of hardware and software to achieve this, including a robust system for estimating the "blur" or loss of detail in downsampled data. Additionally, Cloudflare's analytics APIs provide customers with confidence intervals for their data, allowing them to build dashboards with accurate estimates and confidence bands. Despite the challenges of scaling its data pipeline, Cloudflare continues to work on improving its systems to ensure they remain resilient and useful to its customers.