Back to browse
GitHub Repository

Streaming metrics compression engine. 100M events → 93 KB.

4 starsPython

Sketchlog – 100M events compressed to 93 KB using streaming sketches

by BALAVIGNESH321·Jun 12, 2026·1 point·0 comments

AI Analysis

●●●BangerWizardryBig Brain

100M events to 93 KB with constant memory using probabilistic sketches.

Strengths
  • Genuine algorithmic innovation: constant-memory streaming breaks the volume-to-RAM coupling.
  • Mathematically bounded error rates: DDSketch p99 within 1%, HyperLogLog cardinality within 3%.
  • Dead-simple Python API with no database, backend, or configuration required.
Weaknesses
  • Approximation errors mean it won't work for use cases requiring exact counts or audit trails.
  • Niche appeal: most teams already have Prometheus or Datadog handling this problem.
Category
Target Audience

Backend engineers and SREs handling high-volume metrics

Similar To

Datadog Sketches · RedisBloom · Apache DataSketches

Similar Projects