batch-processing

SKILL.md

30-40% throughput improvement by batching database operations with graceful fallback.

Sequential processing is slow because each item requires multiple DB round trips. The solution is to collect all data first, then execute batch operations:

Key insight: Sequential mapping (fuzzy matching needs context), but batched writes (independent operations).

数据库操作的收集然后批处​​理模式实现了 30-40% 的吞吐量改进。包括在批处理操作失败时优雅地回退到顺序处理。 来源：dadbodgeoff/drift。

查看原文