I would like to get a better understanding of the alternative proposed for elastic-cloud new deployements between "I/O" "CPU" or "memory" optimizations.
What they are, and how the selected hardware is balanced is pretty clear, but the documentation doesn't provide much insight of the use-cases that would better fit each category (except for the high-memory adequation to aggregations).
So I guess my question would be :
- What is a typical "CPU-intensive workload" for elastic ?
- I had the feeling that some of my indices with highly nested mappings run queries better on cpu-intensive than i/o intensive ? Is that accurate ?
- Is there any other case than aggregations (would that include extensive use of kibana ?) for which High-mem is required ?