When we create an SLO, the underlying transforms can potentially triggers some circuit breaker if the source query performs badly. Which can bring the cluster to a halt.
Let's investigate how we could measure the performance of the queries behind a SLO. Then we could think of a way to inform the user about this performance, and suggest a way to mitigate it.
Ideas:
Based on the SLO configuration, generate the composite aggregate query, run it for a small window: look into the duration to execute the search query, and the number of pages yield by the composite.
Get a color status: green, yellow, red indicating performance of the query.
Summary
When we create an SLO, the underlying transforms can potentially triggers some circuit breaker if the source query performs badly. Which can bring the cluster to a halt.
Let's investigate how we could measure the performance of the queries behind a SLO. Then we could think of a way to inform the user about this performance, and suggest a way to mitigate it.
Ideas: