Bucket Scheme Design

Hi, I’d want to know if there are limitations for bucket. I’m a tidy person. Therefore, I want to split metrics by their categories as much as possible.


  • There are 10 dedicated servers. (Content: Memory, CPU, Network traffic etc.)
  • There are 10 web servers. (Content: Visitors, Page load time, Posts, Memory, CPU, Network traffic; BUT for only web server not machine.)

Should I split them by bucket? If the asnwer is yes, I will probably split database and other services.

Technically yes there are. But unless you’re in the extreme minority you will never hit them in practice.

Personally, my current setup is split into a raw/unprocessed bucket with everything in it, and which has a short retention of 1 week.

My processed bucket has various summaries applied to the data and a 1 year retention. Like over a 1min period, I condense all ping points down to a min, median, max. This has benefits in that my dashboards load way quicker, and reduces the storage footprint.

My suggestion is to just get started and experiment. Rather than trying to build a perfect setup on your first try. :+1:

1 Like