Prometheus can easily handle hundreds of thousands up to millions of time series. I recommend you to ignore the number of metrics and focus on the approximate number of time series in your system. Technically even the name of the metric is nothing more than a label called __name__.
>>> services = 100 >>> replicas = 3 >>> endpoints = 20 >>> metric_request_counter = 5 >>> metric_latency = 20 >>> services * replicas * endpoints * metric_request_counter * metric_latency 600000 >>> services * replicas * endpoints * metric_request_counter * 10 300000 Nick schrieb am Mittwoch, 16. September 2020 um 07:59:53 UTC+2: > Thanks guys! I'll try to ensure that all label values are sanitized and > limited to bare minimum, which may still end-up generating around > 2,000-4,000 counter metrics per scrape (retained in Prometheus for up-to a > year). I hope that's not going to be a problem? -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/59cdbe52-26d9-4e0f-a9c4-10cae45359dan%40googlegroups.com.
