Four Hundred Days Of Samples

You are sizing a single Prometheus server. It holds 12,000 active series, scrapes every 15 seconds, and must retain 400 days. Prometheus averages roughly 1.7 bytes per sample after compression, and the documented sizing rule is:

$\text{needed bytes} = \text{retention seconds} \times \text{samples per second} \times \text{bytes per sample}$

Which is closest to the disk required for the retained blocks?