Sampling
To see everything is to see nothing.
The master does not watch every raindrop.
They watch the ones that change the river's course.
In the pipeline you just configured, a processor sits between receiving and exporting. Sampling is one such processor. It decides what to keep.
source
sampler: off
storage
20/20 traces kept.
Apply sampling strategy
Which sampling strategy kept the most useful data with the least volume?
At low traffic you can keep everything. At scale, you can't.
Sampling is the art of choosing what to keep when you can't keep it all.
Continue →