0

Here is my issue, I am looking for a monitoring platform that allows fetching metric from remote server and then digest the metric to create KPIs.

The remote servers are connected to the network through an unreliable connection. Therefore it would need to be able to cache the metrics when the network is down.

On the aggregating server, at each hour, it needs to take all the data from the hours, calculate the KPI and timestamp it as if it was taken at the begginning of the hour. For example, at 13h59 it would timestamp the data as 13h00.

I did some experimentation with prometheus, and it doesn't seems to be the right approach. The experiments, were conducted using prometheus and thanos. Most can be done with those, but I feel that the back in time timestamp is somewhat hackish. It would required using >1hour block size and creating a python script that would fetch the data from the thanos store. Even then, it would work only if I store the KPI in a different database otherwise prometheus will complain that the data are not in chronological order.

thatguy
  • 21,059
  • 6
  • 30
  • 40
Mathieu Tetreault
  • 1

1 Answers1

1

Take a look at VictoriaMetrics. It supports storing historical data. See these docs for more details.

It also provides vmagent tool, which can buffer data on remote servers and flush it to VictoriaMetrics when network is up.

valyala
  • 11,669
  • 1
  • 59
  • 62