/plushcap/analysis/aerospike/aerospike-how-to-spot-a-bad-benchmark

How to Spot a Bad Benchmark

What's this blog post about?

To spot a bad benchmark, consider the following points: 1) Ensure that the physics and measurements involved make sense. Avoid confusing academic terms or misleading marketing claims. 2) Verify if the benchmark provides valuable information about real-world deployments by checking the dataset size and relevance. 3) Be cautious of absurdly overprovisioned benchmarks, as they may not reflect actual usage scenarios. 4) Ensure fairness in benchmark comparisons by avoiding intentional hobbling or unrealistic configurations for competing systems. 5) Check if the benchmark is understandable and reproducible, with clear details on hardware, configurations, and methodologies. Lastly, maintain a critical mindset when evaluating benchmarks.

Company
Aerospike

Date published
June 9, 2020

Author(s)
Ronen Botzer

Word count
609

Hacker News points
None found.

Language
English


By Matt Makai. 2021-2024.