Do you genuinely love log statements, or harbour some bizarre obsession with them!?

NO? Then, you may enjoy reading about the Log Quad: a semantic framework to evaluate your telemetry, and fine-tune your infrastructure bill.

https://www.bahmanm.com/2026/03/log-quad-domesticate-logging-bill.html

#BrainWork_BrainToil #SoftwareEngineering #SRE #O11y #DistributedSystems

@mononcqc I'd very much appreciate any feedback/thoughts.

@bahmanm I read it sort of quickly, these 4 fields make sense for like the key dimensions. If you have trace / request id or some other ID acting as a dimension, it can now work fairly nicely with tail sampling (https://opentelemetry.io/docs/concepts/sampling/#tail-sampling) as a sort of good default way of getting going I think.

That sort of depends on your log intent description as well (eg. intent=audit and you never sample anything, which is different from sampling errors if they all share the same code-marker for example)

Sampling

Learn about sampling and the different sampling options available in OpenTelemetry.

OpenTelemetry