hll experiment (#6312)

## Problem

Measuring cardinality using logs is expensive and slow.

## Summary of changes

Implement a pre-aggregated HyperLogLog-based cardinality estimate.
HyperLogLog estimates the cardinality of a set by using the probability
that the uniform hash of a value will have a run of n 0s at the end is
`1/2^n`, therefore, having observed a run of `n` 0s suggests we have
measured `2^n` distinct values. By using multiple shards, we can use the
harmonic mean to get a more accurate estimate.

We record this into a Prometheus time-series. HyperLogLog counts can be
merged by taking the `max` of each shard. We can apply a `max_over_time`
in order to find the estimate of cardinality of distinct values over
time
This commit is contained in:
Conrad Ludgate
2024-01-29 07:26:20 +00:00
committed by GitHub
parent c1148dc9ac
commit 511e730cc0
8 changed files with 571 additions and 8 deletions
+2 -2
View File
@@ -51,7 +51,7 @@ memchr = { version = "2" }
nom = { version = "7" }
num-bigint = { version = "0.4" }
num-integer = { version = "0.1", features = ["i128"] }
num-traits = { version = "0.2", features = ["i128"] }
num-traits = { version = "0.2", features = ["i128", "libm"] }
once_cell = { version = "1" }
parquet = { git = "https://github.com/neondatabase/arrow-rs", branch = "neon-fix-bugs", default-features = false, features = ["zstd"] }
prost = { version = "0.11" }
@@ -100,7 +100,7 @@ memchr = { version = "2" }
nom = { version = "7" }
num-bigint = { version = "0.4" }
num-integer = { version = "0.1", features = ["i128"] }
num-traits = { version = "0.2", features = ["i128"] }
num-traits = { version = "0.2", features = ["i128", "libm"] }
once_cell = { version = "1" }
parquet = { git = "https://github.com/neondatabase/arrow-rs", branch = "neon-fix-bugs", default-features = false, features = ["zstd"] }
prost = { version = "0.11" }