80% faster, 70% less memory: Go tricks we used to build a high-performance Prometheus query engine

Why watch

This talk presents a case study of building a Prometheus-compatible query engine in Go that achieves 80% faster query execution and 70% lower peak memory usage than Prometheus' default engine. The speakers share hard-won Go performance lessons including when byte slices outperform strings, the practical benefits and implementation costs of memory pooling, and the surprising impact of function pointers on performance. The video is designed for Go developers building high-performance systems who want to learn concrete optimization techniques used in production.