We're looking at a new metrics solution and are attempting to build it in house. So, my question is:
What can scale more effectively / with less pain; Thanos or Cortex?
I understand the general differences between the two. I'm just looking to know if one has some inherent issue at very large scale that the other avoids. E.g. is the Thanos compactor troublesome with lots of time series? (random example). Or do the centralized Cortex pieces have any areas where they don't quite horizontally scale well?
They both sound like they can horizontally scale with almost no limit but that's hard to believe in practice.