pageserver: tweak pageserver_layers_per_read histogram resolution (#10847)

## Problem

The current `pageserver_layers_per_read` histogram buckets don't
represent the current reality very well. For the percentiles we care
about (e.g. p50 and p99), we often see fairly high read amp, especially
during ingestion, and anything below 4 can be considered very good.
 
## Summary of changes

Change the per-timeline read amp histogram buckets to `[4.0, 8.0, 16.0,
32.0, 64.0, 128.0, 256.0]`.
This commit is contained in:
Erik Grinaker
2025-02-17 18:24:17 +01:00
committed by GitHub
parent b10890b81c
commit 84bbe87d60

View File

@@ -130,7 +130,7 @@ pub(crate) static LAYERS_PER_READ: Lazy<HistogramVec> = Lazy::new(|| {
"Layers visited to serve a single read (read amplification). In a batch, all visited layers count towards every read.", "Layers visited to serve a single read (read amplification). In a batch, all visited layers count towards every read.",
&["tenant_id", "shard_id", "timeline_id"], &["tenant_id", "shard_id", "timeline_id"],
// Low resolution to reduce cardinality. // Low resolution to reduce cardinality.
vec![1.0, 5.0, 10.0, 25.0, 50.0, 100.0], vec![4.0, 8.0, 16.0, 32.0, 64.0, 128.0, 256.0],
) )
.expect("failed to define a metric") .expect("failed to define a metric")
}); });