We can create SLOs with metrics (e.g. error budget) which are defined as 1 week, 1 day, or 1 hour.
So, if we choose to configure 1 hour, the error budget will be recredited every hour.
This allows us to obtain a quick view of the state of the service over 1 hour (almost real time in the end).
However, if we display this SLO over a higher time period, say 1 day, or 1 week.
What do the displayed values correspond to? The current 1h values, whatever happens?
Or an averaged value of each hour across the entire selected time period?
Solved! Go to Solution.
That's what I thought but I wasn't sure.
So if I want to build SLOs that are close to a real-time "weather" vision, with error budgets, the best is to create this SLO over 1 hour?
And if C-level management wants from time to time a vision over a higher period, they can position themselves over a 1 week, or more, with values which will be aggregated.
Does my reasoning seem correct to you?