PostgreSQL × Valkey on TPC-H

Reproducible benchmark crossing PostgreSQL indexing with a Valkey in-memory cache tier on analytical TPC-H queries.

How do two independent system-level decisions, relational indexing and an external in-memory cache, interact under analytical workloads? This repository answers that with a 2×2 experimental design on the TPC-H benchmark.

A Rust benchmark harness covering four configurations: indexes on or off × Valkey caching on or off.
Docker Compose brings the whole pipeline up on one workstation.
Ships a full SF10 reference sweep: processed CSVs and figures, interpreted in a companion paper.
Structured IMRaD-style, with documented methods and hardware/software manifests, so anyone can reproduce the results.