Thanks for well written article with many technical details! It was a pleasure to read it!

I have a few questions after reading the article:

* Why did you choose InfluxDB as data storage for Prometheus? It looks like Prometheus itself could be used as data storage.

* Did you consider other solutions for metrics' scraping, storing and processing such as vmagent + VictoriaMetrics? This solution could require less CPU, RAM and disk space comparing to Prometheus+InfluxDB. See https://github.com/VictoriaMetrics/VictoriaMetrics/blob/master/app/vmagent/README.md for details.