Long periods comparison strategy
Use case: World scale e-commerce (100+ microservices, K8S, Lambdas)
Based on basic extrapolation from Q4 2020 data (holiday period), we want to calculate how risky would be to release new changes to our services in the upcoming holidays.
So, the plan is:
- Observe and understand what happened with our services last holiday period (Oct-2020 to Dec-2020).
- Compare periods (ideally monthly) from 2020 vs 2021 for different dashboards. If monthly is too much data, we can compare biweekly, or even weekly.
The question is:
How can we compare information resulted from the same query and same time range but in 2 different years? The Time-compare feature has a 40 days limitation.
For example, I would like to have in a dashboard how many hits I had during August 2021 and August 2020 for some microservices, ideally the same panel, but it would be also OK if we need 2 separated panels. The challenge is to do this at the scale we need (all microservices and Lambdas). I know I could make different dashboards for different parts of the period Oct-2020 Dec-2020, with fixed dates. I'm trying to find a better, replicable solution.
Thanks
Please sign in to leave a comment.
Comments
0 comments