Performance and Scaling

bijux-pollenomics is optimized for reproducibility and inspectability before throughput.

Current Performance Truths

• verification targets are much cheaper than full data refreshes
• source collection cost is dominated by upstream download and normalization work
• report publishing cost grows with the size of tracked source outputs and atlas assets

Scaling Rule

Do not hide performance pressure by collapsing workflow boundaries. If a step is slow, keep the slow step identifiable so reviewers and operators still know whether the cost came from collection, reporting, or docs publication.

Purpose

This page records the package's performance stance and operational trade-offs.